[ { "id": "01-ai-yi-lightning", "name": "yi-lightning", "level": "model", "family_id": null, "developer": [ "01-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "Legacy 01.AI model; no 2026 frontier model found.", "verified_on": "2026-10-10", "sources": [ "https://arena.ai/leaderboard/chat/text" ], "unverified": [] }, { "id": "a-star-meralion-2", "name": "MERaLiON-2", "level": "family", "family_id": null, "developer": [ "a-star" ], "model_class": "multimodal LLM", "modalities_in": [ "audio", "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-05-20", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "meralion-public-license-v3", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English (global and Singapore), Chinese, Singlish, Malay, Tamil, Indonesian, Thai and Vietnamese audio", "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "MERaLiON-2 is a family of speech-text large language models from A*STAR I2R tailored to Singapore and Southeast Asia, with 10B, 10B-ASR and 3B members.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/MERaLiON/MERaLiON-2-10B", "https://leaderboard.sea-lion.ai/detailed/ID" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ], "indonesian_measured": [ { "benchmark": "MERaLiON ASR table, Indonesian row (average word error rate by language)", "score": "0.142813", "metric": "WER (lower is better); same table: whisper_large_v3 0.137102, MERaLiON-2-10B-ASR 0.121020", "run_by": "A*STAR MERaLiON team (developer-reported)", "url": "https://huggingface.co/MERaLiON/MERaLiON-2-10B" }, { "benchmark": "SEA-HELM Indonesian leaderboard (ID aggregate)", "score": "58.01 (95% CI -1.66 to +1.65)", "metric": "score 0 to 100, average of 2000 bootstraps with 95% CI", "run_by": "AI Singapore (SEA-HELM leaderboard maintainer)", "url": "https://leaderboard.sea-lion.ai/detailed/ID" } ] }, { "id": "a-star-meralion-2-10b", "name": "MERaLiON-2-10B", "level": "model", "family_id": "a-star-meralion-2", "developer": [ "a-star" ], "model_class": "multimodal LLM", "modalities_in": [ "audio", "text" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "Localised Whisper-large-v3 speech encoder with a Gemma2-9b-IT text decoder", "params_total": 10106437888, "params_active": null, "params_checkpoint": null, "params_note": "Hugging Face API Safetensors count", "context_length": null, "release_date": "2025-05-20", "release_date_basis": "registry date", "weights_date": "2025-05-20", "openness": "open weights restricted", "licence": "meralion-public-license-v3", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": "Audio input up to 300 seconds; 30 seconds recommended for ASR and speech translation", "runtimes": [ "vllm" ], "languages_note": "Primarily English (global and Singapore) and Chinese, with audio support for Malay, Tamil, Indonesian, Thai and Vietnamese", "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MERaLiON-2-10B is an A*STAR speech-text LLM for ASR, speech translation, spoken QA and audio understanding tuned to Singapore and Southeast Asian languages, trained on 120,000 hours of audio.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/MERaLiON/MERaLiON-2-10B", "https://huggingface.co/api/models/MERaLiON/MERaLiON-2-10B", "https://leaderboard.sea-lion.ai/detailed/ID" ], "unverified": [ "indonesia_hosting" ], "indonesian_measured": [ { "benchmark": "MERaLiON ASR table, Indonesian row (average word error rate by language)", "score": "0.142813", "metric": "WER (lower is better); same table: whisper_large_v3 0.137102, MERaLiON-2-10B-ASR 0.121020", "run_by": "A*STAR MERaLiON team (developer-reported)", "url": "https://huggingface.co/MERaLiON/MERaLiON-2-10B" }, { "benchmark": "SEA-HELM Indonesian leaderboard (ID aggregate)", "score": "58.01 (95% CI -1.66 to +1.65)", "metric": "score 0 to 100, average of 2000 bootstraps with 95% CI", "run_by": "AI Singapore (SEA-HELM leaderboard maintainer)", "url": "https://leaderboard.sea-lion.ai/detailed/ID" } ] }, { "id": "a-star-meralion-2-3b", "name": "MERaLiON-2-3B", "level": "model", "family_id": "a-star-meralion-2", "developer": [ "a-star" ], "model_class": "multimodal LLM", "modalities_in": [ "audio", "text" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "Localised Whisper-large-v3 speech encoder with a Gemma2-2b-IT text decoder", "params_total": 3470880512, "params_active": null, "params_checkpoint": null, "params_note": "Hugging Face API Safetensors count", "context_length": null, "release_date": "2025-05-25", "release_date_basis": "registry date", "weights_date": "2025-05-25", "openness": "open weights restricted", "licence": "meralion-public-license-v3", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [ "vllm" ], "languages_note": "Primarily English (global and Singapore) and Chinese, with audio support for Malay, Tamil, Indonesian, Thai and Vietnamese", "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MERaLiON-2-3B is the lightweight member of the A*STAR MERaLiON-2 speech-text LLM family, balancing performance with lower compute.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/MERaLiON/MERaLiON-2-3B", "https://huggingface.co/MERaLiON/MERaLiON-2-10B", "https://huggingface.co/api/models/MERaLiON/MERaLiON-2-3B" ], "unverified": [ "indonesia_hosting" ], "indonesian_measured": [ { "benchmark": "MERaLiON ASR table, Indonesian row (average word error rate by language)", "score": "0.131950", "metric": "WER (lower is better); same table: whisper_large_v3 0.137102, MERaLiON-2-10B-ASR 0.121020", "run_by": "A*STAR MERaLiON team (developer-reported)", "url": "https://huggingface.co/MERaLiON/MERaLiON-2-10B" } ] }, { "id": "abot-world-v0-1", "name": "ABot-World v0.1", "level": "model", "family_id": null, "developer": [], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Video model listed on the VBench leaderboard; developer not stated in the notes.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/spaces/Vchitect/VBench_Leaderboard" ], "unverified": [] }, { "id": "absolute-zero", "name": "Absolute Zero", "level": "model", "family_id": null, "developer": [ "tsinghua-university" ], "model_class": "reasoning", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-05", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Self-play reasoning method and model using a code executor with no external data; compute not reported; developer not stated in the notes.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2505.03335", "https://arxiv.org/html/2505.03335" ], "unverified": [] }, { "id": "adobe-firefly-image", "name": "Firefly Image", "level": "family", "family_id": null, "developer": [ "adobe" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Adobe Firefly image model line.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Adobe_Firefly" ], "unverified": [] }, { "id": "adobe-firefly-image-4", "name": "Firefly Image 4", "level": "model", "family_id": "adobe-firefly-image", "developer": [ "adobe" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-04", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Adobe Firefly Image 4 and 4 Ultra (Apr 2025).", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Adobe_Firefly" ], "unverified": [] }, { "id": "adobe-firefly-image-5", "name": "Firefly Image 5", "level": "model", "family_id": "adobe-firefly-image", "developer": [ "adobe" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Adobe image model in the Firefly app and Creative Cloud, marketed as commercially safe (Adobe Stock training); C2PA manifests on Firefly outputs per a secondary source.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Adobe_Firefly", "https://c2paviewer.com/articles/ai-tools-c2pa-support", "https://en.wikipedia.org/wiki/Adobe_Inc." ], "unverified": [] }, { "id": "agibot-go", "name": "GO", "level": "family", "family_id": null, "developer": [ "agibot" ], "model_class": "robot policy", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "AgiBot generalist robot foundation models.", "verified_on": "2026-10-10", "sources": [ "https://www.therobotreport.com/agibot-releases-go-2-foundation-model-embodied-ai/" ], "unverified": [] }, { "id": "agibot-go-1", "name": "GO-1", "level": "model", "family_id": "agibot-go", "developer": [ "agibot" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": null, "architecture_detail": "ViLLA (vision-language-latent-action)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-04", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "AgiBot VLA released about a year before GO-2.", "verified_on": "2026-10-10", "sources": [ "https://www.therobotreport.com/agibot-releases-go-2-foundation-model-embodied-ai/" ], "unverified": [] }, { "id": "agibot-go-2", "name": "GO-2", "level": "model", "family_id": "agibot-go", "developer": [ "agibot" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": null, "architecture_detail": "Action chain-of-thought plus asynchronous dual system (low-frequency planner, high-frequency residual executor)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04-09", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "AgiBot foundation model with vendor-reported 98.5% LIBERO average; openness not stated.", "verified_on": "2026-10-10", "sources": [ "https://www.therobotreport.com/agibot-releases-go-2-foundation-model-embodied-ai/" ], "unverified": [] }, { "id": "ai-singapore-gemma-sea-guard-12b-2602", "name": "Gemma-SEA-Guard-12B-2602", "level": "model", "family_id": "ai-singapore-sea-guard", "developer": [ "ai-singapore" ], "model_class": "guard", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 12000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-02-02", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": "English, Burmese, Indonesian, Malay, Tagalog, Tamil, Thai, Vietnamese", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Gemma-SEA-Guard-12B-2602, an AI Singapore SEA-Guard binary safe/unsafe classifier; benchmarked on SEA-SafeguardBench (AUPRC, scores only in an image).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/aisingapore", "https://huggingface.co/aisingapore/Qwen-SEA-Guard-8B-2602" ], "unverified": [ "licence", "indonesian_measured", "indonesia_hosting" ] }, { "id": "ai-singapore-gemma-sea-lion-v4-27b", "name": "Gemma-SEA-LION-v4-27B", "level": "model", "family_id": null, "developer": [ "ai-singapore" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "27B", "context_length": null, "release_date": "2025-08-11", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "AI Singapore SEA-LION v4 27B, finetuned from gemma-3-27b-it.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/api/models/aisingapore/Gemma-SEA-LION-v4-27B", "https://huggingface.co/aisingapore/Gemma-SEA-LION-v4-27B" ], "unverified": [] }, { "id": "ai-singapore-gemma-sea-lion-v4-27b-it", "name": "Gemma-SEA-LION-v4-27B-IT", "level": "model", "family_id": null, "developer": [ "ai-singapore" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "27B", "context_length": null, "release_date": "2025-08-11", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Instruct version of SEA-LION v4 27B.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/api/models/aisingapore/Gemma-SEA-LION-v4-27B", "https://huggingface.co/aisingapore/Gemma-SEA-LION-v4-27B-IT" ], "unverified": [] }, { "id": "ai-singapore-gemma-sea-lion-v4-5-e2b-it", "name": "Gemma-SEA-LION-v4.5-E2B-IT", "level": "model", "family_id": "ai-singapore-sea-lion", "developer": [ "ai-singapore" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 5100000000, "params_active": 2300000000, "params_checkpoint": null, "params_note": "2.3B effective (5.1B total)", "context_length": null, "release_date": "2026-05-19", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "laptop", "mobile or NPU" ], "footprint_note": "GGUF", "runtimes": [], "languages_note": "9 languages including Indonesian", "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "AI Singapore's small edge SEA model from gemma-4-E2B-it, distilled from gemma-4-31B-it plus merging.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/aisingapore/Gemma-SEA-LION-v4.5-E2B-IT", "https://leaderboard.sea-lion.ai/detailed/SEA", "https://leaderboard.sea-lion.ai/detailed/ID", "https://huggingface.co/aisingapore" ], "unverified": [ "indonesia_hosting" ], "indonesian_measured": [ { "benchmark": "SEA-HELM Indonesian leaderboard (ID aggregate, first score column of the Indonesian Competency and Task Scores table)", "score": "63.96 (95% CI -1.38 to +1.37)", "metric": "score 0 to 100, average of 2000 bootstraps with 95% CI", "run_by": "AI Singapore (SEA-HELM leaderboard maintainer)", "url": "https://leaderboard.sea-lion.ai/detailed/ID" } ] }, { "id": "ai-singapore-gemma2-9b-cpt-sea-lionv3-base", "name": "gemma2-9b-cpt-sea-lionv3-base (SEA-LION v3)", "level": "model", "family_id": null, "developer": [ "ai-singapore" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "9B (per name)", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "AI Singapore SEA-LION v3 continued-pretraining base on Gemma 2; repo metadata not read.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/api/models/GoToCompany/gemma2-9b-cpt-sahabatai-v1-base" ], "unverified": [ "params_note" ] }, { "id": "ai-singapore-llama-sea-guard-8b-2602", "name": "Llama-SEA-Guard-8B-2602", "level": "model", "family_id": "ai-singapore-sea-guard", "developer": [ "ai-singapore" ], "model_class": "guard", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-02-02", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": "English, Burmese, Indonesian, Malay, Tagalog, Tamil, Thai, Vietnamese", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Llama-SEA-Guard-8B-2602, an AI Singapore SEA-Guard binary safe/unsafe classifier; benchmarked on SEA-SafeguardBench (AUPRC, scores only in an image).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/aisingapore", "https://huggingface.co/aisingapore/Qwen-SEA-Guard-8B-2602" ], "unverified": [ "licence", "indonesian_measured", "indonesia_hosting" ] }, { "id": "ai-singapore-nemotron-sea-lion-v4-8-120b-a12b", "name": "Nemotron-SEA-LION-v4.8-120B-A12B", "level": "model", "family_id": "ai-singapore-sea-lion", "developer": [ "ai-singapore" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Mamba2-Transformer hybrid MoE", "params_total": 124000000000, "params_active": 12000000000, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-09-18", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": "Single GPU at NVFP4 (64B)", "runtimes": [], "languages_note": "English + 7 SEA languages including Indonesian", "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "AI Singapore's Nemotron-based SEA model (base and instruct); tech report arXiv 2609.18310.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/aisingapore/Nemotron-SEA-LION-v4.8-120B-A12B", "https://huggingface.co/models?search=sea-lion%20v4.8", "https://leaderboard.sea-lion.ai/detailed/SEA", "https://leaderboard.sea-lion.ai/detailed/ID" ], "unverified": [ "indonesia_hosting" ], "indonesian_measured": [ { "benchmark": "SEA-HELM Indonesian leaderboard (ID aggregate, first score column of the Indonesian Competency and Task Scores table)", "score": "73.10 (95% CI -1.16 to +1.21)", "metric": "score 0 to 100, average of 2000 bootstraps with 95% CI", "run_by": "AI Singapore (SEA-HELM leaderboard maintainer)", "url": "https://leaderboard.sea-lion.ai/detailed/ID" } ] }, { "id": "ai-singapore-nemotron-sea-lion-v4-8-30b-a3b", "name": "Nemotron-SEA-LION-v4.8-30B-A3B", "level": "model", "family_id": "ai-singapore-sea-lion", "developer": [ "ai-singapore" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Mamba2-Transformer hybrid MoE", "params_total": 33000000000, "params_active": 3000000000, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-09-18", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": "Consumer GPU at NVFP4 (17B)", "runtimes": [], "languages_note": "English + 7 SEA languages including Indonesian", "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "AI Singapore's Nemotron-based SEA model (base and instruct); tech report arXiv 2609.18310.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/aisingapore/Nemotron-SEA-LION-v4.8-120B-A12B", "https://huggingface.co/models?search=sea-lion%20v4.8", "https://leaderboard.sea-lion.ai/detailed/SEA", "https://leaderboard.sea-lion.ai/detailed/ID" ], "unverified": [ "indonesia_hosting" ], "indonesian_measured": [ { "benchmark": "SEA-HELM Indonesian leaderboard (ID aggregate, first score column of the Indonesian Competency and Task Scores table)", "score": "65.87 (95% CI -1.29 to +1.33)", "metric": "score 0 to 100, average of 2000 bootstraps with 95% CI", "run_by": "AI Singapore (SEA-HELM leaderboard maintainer)", "url": "https://leaderboard.sea-lion.ai/detailed/ID" } ] }, { "id": "ai-singapore-qwen-sea-guard-8b-2602", "name": "Qwen-SEA-Guard-8B-2602", "level": "model", "family_id": "ai-singapore-sea-guard", "developer": [ "ai-singapore" ], "model_class": "guard", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-02-02", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": "English, Burmese, Indonesian, Malay, Tagalog, Tamil, Thai, Vietnamese", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen-SEA-Guard-8B-2602, an AI Singapore SEA-Guard binary safe/unsafe classifier; benchmarked on SEA-SafeguardBench (AUPRC, scores only in an image).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/aisingapore", "https://huggingface.co/aisingapore/Qwen-SEA-Guard-8B-2602" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "ai-singapore-qwen-sea-lion-v4-5-27b-it", "name": "Qwen-SEA-LION-v4.5-27B-IT", "level": "model", "family_id": "ai-singapore-sea-lion", "developer": [ "ai-singapore" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 27000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-05-19", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": "OpenVINO 4- and 8-bit variants (v4-32B)", "runtimes": [ "openvino" ], "languages_note": "English, Burmese, Indonesian, Filipino, Malay, Tamil, Thai, Vietnamese", "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "AI Singapore's 27B Southeast Asian model post-trained on Qwen3.6-27B by logit distillation and merging on SEA-Instruct-2602.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/aisingapore/Qwen-SEA-LION-v4.5-27B-IT", "https://leaderboard.sea-lion.ai/detailed/SEA", "https://leaderboard.sea-lion.ai/detailed/ID" ], "unverified": [ "indonesia_hosting" ], "indonesian_measured": [ { "benchmark": "SEA-HELM Indonesian leaderboard (ID aggregate, first score column of the Indonesian Competency and Task Scores table)", "score": "76.51 (95% CI -1.23 to +1.23)", "metric": "score 0 to 100, average of 2000 bootstraps with 95% CI", "run_by": "AI Singapore (SEA-HELM leaderboard maintainer)", "url": "https://leaderboard.sea-lion.ai/detailed/ID" } ] }, { "id": "ai-singapore-qwen-sea-lion-v4-8b-vl", "name": "Qwen-SEA-LION-v4-8B-VL", "level": "model", "family_id": "ai-singapore-sea-lion", "developer": [ "ai-singapore" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-11-21", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/aisingapore/Qwen-SEA-Guard-8B-2602", "https://huggingface.co/aisingapore/Qwen-SEA-LION-v4-8B-VL" ], "unverified": [] }, { "id": "ai-singapore-sea-guard", "name": "SEA-Guard", "level": "family", "family_id": null, "developer": [ "ai-singapore" ], "model_class": "guard", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-02-02", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English, Burmese, Indonesian, Malay, Tagalog, Tamil, Thai, Vietnamese", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "AI Singapore's guard models giving binary safe/unsafe output for text and vision; the only verified open guard model listing Indonesian (paper arXiv 2602.01618). Card licence is inconsistent (\"other\" in metadata, Apache-2.0 in text).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/aisingapore", "https://huggingface.co/aisingapore/Qwen-SEA-Guard-8B-2602" ], "unverified": [ "licence", "indonesian_measured", "indonesia_hosting" ] }, { "id": "ai-singapore-sea-lion", "name": "SEA-LION", "level": "family", "family_id": null, "developer": [ "ai-singapore" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "AI Singapore's SEA-LION family of Southeast Asian language models.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/aisingapore", "https://huggingface.co/aisingapore/Qwen-SEA-LION-v4.5-27B-IT", "https://leaderboard.sea-lion.ai/detailed/ID", "https://huggingface.co/aisingapore/Gemma-SEA-LION-v4.5-E2B-IT" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ], "indonesian_measured": [ { "benchmark": "SEA-HELM Indonesian leaderboard (ID aggregate, first score column of the Indonesian Competency and Task Scores table)", "score": "76.51 (95% CI -1.23 to +1.23)", "metric": "score 0 to 100, average of 2000 bootstraps with 95% CI", "run_by": "AI Singapore (SEA-HELM leaderboard maintainer)", "url": "https://leaderboard.sea-lion.ai/detailed/ID" } ] }, { "id": "ai-singapore-sea-lion-e5-embedding-600m", "name": "SEA-LION-E5-Embedding-600M", "level": "model", "family_id": "ai-singapore-sea-lion", "developer": [ "ai-singapore" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "E5-style embedding", "params_total": 600000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-10", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": null, "runtimes": [], "languages_note": "11 languages: Burmese, Chinese, English, Filipino, Indonesian, Khmer, Lao, Malay, Tamil, Thai, Vietnamese", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "AI Singapore SEA-LION-E5-Embedding-600M encoder or embedding model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/aisingapore", "https://huggingface.co/aisingapore/SEA-LION-E5-Embedding-600M" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "ai-singapore-sea-lion-modernbert-300m", "name": "SEA-LION ModernBERT 300M", "level": "model", "family_id": "ai-singapore-sea-lion", "developer": [ "ai-singapore" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "ModernBERT encoder", "params_total": 300000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-10", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": null, "runtimes": [], "languages_note": "13 languages plus code: Burmese, Chinese, English, Filipino, Indonesian, Javanese, Khmer, Lao, Malay, Sundanese, Tamil, Thai, Vietnamese", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "AI Singapore SEA-LION ModernBERT 300M encoder or embedding model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/aisingapore", "https://huggingface.co/aisingapore/SEA-LION-ModernBERT-300M" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "ai-singapore-sea-lion-modernbert-600m", "name": "SEA-LION ModernBERT 600M", "level": "model", "family_id": "ai-singapore-sea-lion", "developer": [ "ai-singapore" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "ModernBERT encoder", "params_total": 600000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-10", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": null, "runtimes": [], "languages_note": "13 languages plus code: Burmese, Chinese, English, Filipino, Indonesian, Javanese, Khmer, Lao, Malay, Sundanese, Tamil, Thai, Vietnamese", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "AI Singapore SEA-LION ModernBERT 600M encoder or embedding model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/aisingapore", "https://huggingface.co/aisingapore/SEA-LION-ModernBERT-600M" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "ai-singapore-sea-lion-modernbert-embedding-300m", "name": "SEA-LION ModernBERT-Embedding 300M", "level": "model", "family_id": "ai-singapore-sea-lion", "developer": [ "ai-singapore" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "ModernBERT encoder", "params_total": 300000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-10", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "CPU server" ], "footprint_note": null, "runtimes": [], "languages_note": "SEA languages incl. Indonesian (unverified list)", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "AI Singapore SEA-LION ModernBERT-Embedding 300M encoder or embedding model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/aisingapore" ], "unverified": [ "licence", "languages_note" ] }, { "id": "ai-singapore-sea-lion-modernbert-embedding-600m", "name": "SEA-LION ModernBERT-Embedding 600M", "level": "model", "family_id": "ai-singapore-sea-lion", "developer": [ "ai-singapore" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "ModernBERT encoder", "params_total": 600000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-10", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "CPU server" ], "footprint_note": null, "runtimes": [], "languages_note": "SEA languages incl. Indonesian (unverified list)", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "AI Singapore SEA-LION ModernBERT-Embedding 600M encoder or embedding model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/aisingapore" ], "unverified": [ "licence", "languages_note" ] }, { "id": "ai-singapore-sea-lion-v3", "name": "SEA-LION v3", "level": "model", "family_id": null, "developer": [ "ai-singapore" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "8B Llama-based and 9B Gemma-based variants; built with large-scale multilingual continued pretraining.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "11 Southeast Asian languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Southeast Asian language model built by continued pretraining of 8B and 9B open models; AI Singapore affiliation not confirmed in the page read.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2504.05747" ], "unverified": [] }, { "id": "ai-singapore-sea-lion-v4-5-qwen-27b", "name": "SEA-LION v4.5 (Qwen) 27B", "level": "model", "family_id": null, "developer": [ "ai-singapore" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "27B; Qwen-based per leaderboard name, repo not checked", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "AI Singapore SEA-LION v4.5 27B on a Qwen base, ranked fourth on SEA-HELM Indonesian.", "verified_on": "2026-10-10", "sources": [ "https://leaderboard.sea-lion.ai/detailed/ID", "https://huggingface.co/aisingapore/Qwen-SEA-LION-v4.5-27B-IT" ], "unverified": [ "indonesia_hosting" ], "indonesian_measured": [ { "benchmark": "SEA-HELM Indonesian leaderboard (ID aggregate, first score column of the Indonesian Competency and Task Scores table)", "score": "76.51 (95% CI -1.23 to +1.23)", "metric": "score 0 to 100, average of 2000 bootstraps with 95% CI", "run_by": "AI Singapore (SEA-HELM leaderboard maintainer)", "url": "https://leaderboard.sea-lion.ai/detailed/ID" } ] }, { "id": "ai21-labs-jamba-1-5", "name": "Jamba 1.5", "level": "family", "family_id": null, "developer": [ "ai21-labs" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-08-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "jamba-open-model-license", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "AI21 Labs Jamba 1.5 open model family of Jamba 1.5 Mini and Jamba 1.5 Large.", "verified_on": "2026-10-11", "sources": [ "https://www.ai21.com/blog/announcing-jamba-model-family/" ], "unverified": [] }, { "id": "ai21-labs-jamba-1-5-large", "name": "Jamba 1.5 Large", "level": "model", "family_id": "ai21-labs-jamba-1-5", "developer": [ "ai21-labs" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "SSM-Transformer (Mamba plus Transformer) with MoE layers", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 256000, "release_date": "2024-08-22", "release_date_basis": "announcement page", "weights_date": "2024-08-22", "openness": "open weights restricted", "licence": "jamba-open-model-license", "access_routes": [ "API", "cloud marketplace", "open weights download" ], "deployment_tiers": [ "multi-GPU node" ], "footprint_note": "Cannot be loaded in FP32 or BF16 on a single node of 8 GPUs; AI21 ExpertsInt8 quantization lets it fit on one 8-GPU node", "runtimes": [], "languages_note": "English, Spanish, French, Portuguese, Italian, Dutch, German, Arabic, Hebrew", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "AI21 Labs open long-context model on the hybrid SSM-Transformer Jamba architecture with a 256K context window, aimed at enterprise document and RAG workloads.", "verified_on": "2026-10-11", "sources": [ "https://www.ai21.com/blog/announcing-jamba-model-family/", "https://www.ai21.com/jamba-open-model-license" ], "unverified": [ "status" ] }, { "id": "ai21-labs-jamba-1-5-mini", "name": "Jamba 1.5 Mini", "level": "model", "family_id": null, "developer": [ "ai21-labs" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Hybrid SSM-Transformer (joint attention and Mamba) with mixture of experts", "params_total": 52000000000, "params_active": 12000000000, "params_checkpoint": null, "params_note": null, "context_length": 256000, "release_date": "2024-08-22", "release_date_basis": "announcement page", "weights_date": "2024-08-19", "openness": "open weights restricted", "licence": "jamba-open-model-license", "access_routes": [ "API", "cloud marketplace", "open weights download" ], "deployment_tiers": [ "multi-GPU node", "single datacentre GPU" ], "footprint_note": "BF16 needs at least 2 x 80GB GPUs; ExpertsInt8 in vLLM fits a single 80GB GPU with prompts up to about 100K tokens; blog says up to 140K tokens on a single GPU", "runtimes": [ "vllm", "sglang", "llama-cpp", "ollama" ], "languages_note": "English, Spanish, French, Portuguese, Italian, Dutch, German, Arabic, Hebrew", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Jamba 1.5 Mini is an AI21 hybrid SSM-Transformer mixture-of-experts model with a 256K context window, 12B active of 52B total parameters.", "verified_on": "2026-10-11", "sources": [ "https://www.ai21.com/blog/announcing-jamba-model-family", "https://huggingface.co/ai21labs/AI21-Jamba-Mini-1.5", "https://huggingface.co/api/models/ai21labs/AI21-Jamba-1.5-Mini" ], "unverified": [] }, { "id": "ai21-labs-jamba-1-7-large", "name": "Jamba 1.7 Large", "level": "model", "family_id": null, "developer": [ "ai21-labs" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 256000, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/leaderboards/models", "https://www.ai21.com/blog/" ], "unverified": [] }, { "id": "ai4bharat-indic-parler-tts", "name": "ai4bharat indic-parler-tts", "level": "model", "family_id": null, "developer": [ "ai4bharat" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-10-28", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open TTS model trending on Hugging Face (2025 to 2026).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/models?pipeline_tag=text-to-speech&sort=trending", "https://huggingface.co/ai4bharat/indic-parler-tts" ], "unverified": [] }, { "id": "aircas-ringmo", "name": "RingMo family", "level": "family", "family_id": null, "developer": [ "aircas" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "AIRCAS remote sensing foundation model family.", "verified_on": "2026-10-10", "sources": [ "https://english.cas.cn/newsroom/research_news/infotech/202209/t20220909_320270.shtml", "https://arxiv.org/abs/2504.03166" ], "unverified": [] }, { "id": "aircas-ringmo-aerial", "name": "RingMo-Aerial", "level": "model", "family_id": "aircas-ringmo", "developer": [ "aircas" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "other", "architecture_detail": "Affine-transformation contrastive learning for aerial imagery", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-09-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "RingMo model for aerial imagery.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2409.13366" ], "unverified": [ "developer" ] }, { "id": "aircas-ringmo-agent", "name": "RingMo-Agent", "level": "model", "family_id": "aircas-ringmo", "developer": [ "aircas" ], "model_class": "geospatial", "modalities_in": [ "image", "text" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "Multi-platform, multi-modal perception and reasoning from text instructions; RS-VL3M dataset", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-07-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "RingMo vision-language model.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2507.20776" ], "unverified": [] }, { "id": "aircas-ringmo-lite", "name": "RingMo-lite", "level": "model", "family_id": "aircas-ringmo", "developer": [ "aircas" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "hybrid", "architecture_detail": "CNN-Transformer hybrid, lightweight multi-task", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-09-16", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "embedded GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Lightweight RingMo model.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2309.09003" ], "unverified": [ "developer" ] }, { "id": "aircas-ringmoe", "name": "RingMoE", "level": "model", "family_id": "aircas-ringmo", "developer": [ "aircas" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "MoE", "architecture_detail": "Hierarchical mixture of modality experts over optical, SAR and multispectral", "params_total": 14700000000, "params_active": null, "params_checkpoint": null, "params_note": "Pre-trained on 400M multi-modal images from nine satellites", "context_length": null, "release_date": "2025-04-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "hyperscale cluster" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "AIRCAS 14.7B multimodal remote sensing model.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2504.03166" ], "unverified": [] }, { "id": "alibaba-cam-plus-plus", "name": "CAM++", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "speaker", "modalities_in": [ "audio" ], "modalities_out": [], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights restricted", "licence": "funasr-model-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Speaker embedding component in FunASR.", "verified_on": "2026-10-10", "sources": [ "https://github.com/modelscope/FunASR" ], "unverified": [] }, { "id": "alibaba-damo-seallms-audio-7b", "name": "SeaLLMs-Audio-7B", "level": "model", "family_id": null, "developer": [ "alibaba-damo" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "audio" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 7000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-12", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/SeaLLMs" ], "unverified": [ "licence" ] }, { "id": "alibaba-damo-seallms-v3", "name": "SeaLLMs v3", "level": "family", "family_id": null, "developer": [ "alibaba-damo" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-07", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "consumer GPU", "CPU server" ], "footprint_note": null, "runtimes": [], "languages_note": "SEA languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba DAMO's SEA language models (1.5B, 7B); no 2026 text release found.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/SeaLLMs" ], "unverified": [ "licence" ] }, { "id": "alibaba-damo-seallms-v3-1-5b", "name": "SeaLLMs v3 1.5B", "level": "model", "family_id": "alibaba-damo-seallms-v3", "developer": [ "alibaba-damo" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 1500000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-07", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/SeaLLMs" ], "unverified": [ "licence" ] }, { "id": "alibaba-damo-seallms-v3-7b", "name": "SeaLLMs v3 7B", "level": "model", "family_id": "alibaba-damo-seallms-v3", "developer": [ "alibaba-damo" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 7000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-07", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/SeaLLMs" ], "unverified": [ "licence" ] }, { "id": "alibaba-fun-asr-mlt-nano", "name": "Fun-ASR-MLT-Nano", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": 800000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": "31 languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multilingual Fun-ASR-Nano version (31 languages).", "verified_on": "2026-10-10", "sources": [ "https://github.com/modelscope/FunASR" ], "unverified": [ "licence" ] }, { "id": "alibaba-fun-asr-nano", "name": "Fun-ASR-Nano (Fun-ASR-Nano-2512)", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": 800000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-12-15", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [ "llama-cpp" ], "languages_note": "zh/en/ja + Chinese dialects", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba FunAudioLLM ASR model. Licence resolved to Apache-2.0: no licence file in the repos; model card and publisher GitHub state Apache-2.0.", "verified_on": "2026-10-10", "sources": [ "https://github.com/modelscope/FunASR", "https://huggingface.co/FunAudioLLM/Fun-ASR-Nano-2512", "https://github.com/FunAudioLLM/Fun-ASR", "https://github.com/modelscope/FunASR", "https://huggingface.co/api/models/FunAudioLLM/Fun-ASR-Nano-2512" ], "unverified": [ "architecture" ] }, { "id": "alibaba-gte-multilingual-base", "name": "gte-multilingual-base", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [], "architecture": "other", "architecture_detail": "Encoder-only transformer; elastic dense embeddings (128 to 768) and sparse vectors", "params_total": 305000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 8192, "release_date": "2024-07-20", "release_date_basis": "registry date", "weights_date": "2024-07-20", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Over 70 languages; card language tags include Indonesian", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "gte-multilingual-base is an Alibaba GTE text embedding model for multilingual retrieval with 8192-token input, dense and sparse outputs.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/Alibaba-NLP/gte-multilingual-base", "https://huggingface.co/api/models/Alibaba-NLP/gte-multilingual-base" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "alibaba-happy-oyster", "name": "Happy Oyster", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "world model", "modalities_in": [], "modalities_out": [ "3D", "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04-16", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [ "gated preview" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba ATH AI Innovation Unit real-time interactive 3D world creation model with a Directing mode; not open source, early-access waitlist.", "verified_on": "2026-10-10", "sources": [ "https://blog.buildfastwithai.com/tencent-alibaba-world-models-april-2026" ], "unverified": [] }, { "id": "alibaba-happyhorse-1-0", "name": "HappyHorse 1.0", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": "Developer listed as Alibaba-ATH", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba-ATH proprietary video model.", "verified_on": "2026-10-10", "sources": [ "https://arena.ai/leaderboard/text-to-video" ], "unverified": [] }, { "id": "alibaba-happyhorse-1-1", "name": "HappyHorse 1.1", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": "Developer listed as Alibaba-ATH", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-06", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba-ATH proprietary video model.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/video/leaderboard/text-to-video", "https://arena.ai/leaderboard/text-to-video" ], "unverified": [ "developer" ] }, { "id": "alibaba-lingjiang-2-0", "name": "Lingjiang2.0", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba Cloud zero-shot forecasting model on GIFT-Eval.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/spaces/Salesforce/GIFT-Eval" ], "unverified": [] }, { "id": "alibaba-paraformer-zh", "name": "Paraformer-zh (+ streaming)", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "Non-autoregressive Paraformer", "params_total": 220000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "funasr-model-license", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server", "embedded GPU" ], "footprint_note": null, "runtimes": [], "languages_note": "Chinese/English with timestamps", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba DAMO/Tongyi non-autoregressive ASR model.", "verified_on": "2026-10-10", "sources": [ "https://github.com/modelscope/FunASR" ], "unverified": [ "release_date" ] }, { "id": "alibaba-qwen", "name": "Qwen", "level": "family", "family_id": null, "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen LLM; supplies the decoder of Canary-Qwen-2.5B.", "verified_on": "2026-10-10", "sources": [ "https://www.gladia.io/blog/best-open-source-speech-to-text-models" ], "unverified": [] }, { "id": "alibaba-qwen-14b-chat", "name": "Qwen-14B-Chat", "level": "model", "family_id": "alibaba-qwen", "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 14000000000, "params_active": null, "params_checkpoint": null, "params_note": "Hugging Face API counts 14,167,290,880", "context_length": null, "release_date": "2023-09-24", "release_date_basis": "registry date", "weights_date": "2023-09-24", "openness": "open weights restricted", "licence": "tongyi-qianwen-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Card language tags: zh, en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Qwen-14B-Chat is the aligned chat assistant built on Qwen-14B, the 14B first-generation Tongyi Qianwen model from Alibaba Cloud; commercial use is by application form.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/Qwen/Qwen-14B-Chat", "https://huggingface.co/api/models/Qwen/Qwen-14B-Chat" ], "unverified": [] }, { "id": "alibaba-qwen-audio-3-1-tts-plus", "name": "Qwen-Audio-3.1-TTS-Plus", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-09", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "TTS model 3rd on the Artificial Analysis TTS arena; Alibaba attribution inferred from the name.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/text-to-speech/leaderboard" ], "unverified": [ "developer", "access_routes" ] }, { "id": "alibaba-qwen-image", "name": "Qwen-Image", "level": "model", "family_id": "alibaba-qwen-image-family", "developer": [ "alibaba" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": 20000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-08", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Original Qwen-Image model, Aug 2025, Apache 2.0, 20B.", "verified_on": "2026-10-10", "sources": [ "https://datanorth.ai/news/qwen-releases-qwen-image-2-1", "https://huggingface.co/Qwen/Qwen-Image-2.1" ], "unverified": [] }, { "id": "alibaba-qwen-image-2-1", "name": "Qwen-Image-2.1", "level": "model", "family_id": "alibaba-qwen-image-family", "developer": [ "alibaba" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": "Single-Stream DiT, 32 layers; Qwen3-VL 8B text encoder", "params_total": 7000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-09-20", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights restricted", "licence": "qwen-research-license", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": "Single GPU with CPU offload; VRAM not published", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Top open-weights image model on Artificial Analysis, non-commercial Qwen Research License; heavy Hugging Face ecosystem of Turbo, GGUF and LoRA derivatives.", "verified_on": "2026-10-10", "sources": [ "https://datanorth.ai/news/qwen-releases-qwen-image-2-1", "https://huggingface.co/Qwen/Qwen-Image-2.1", "https://huggingface.co/models?pipeline_tag=text-to-image&sort=trending", "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text", "https://arena.ai/leaderboard/text-to-image" ], "unverified": [] }, { "id": "alibaba-qwen-image-2512", "name": "Qwen-Image-2512", "level": "model", "family_id": "alibaba-qwen-image-family", "developer": [ "alibaba" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-12-30", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen-Image release merged into Qwen-Image-2.1.", "verified_on": "2026-10-10", "sources": [ "https://datanorth.ai/news/qwen-releases-qwen-image-2-1", "https://huggingface.co/Qwen/Qwen-Image-2512" ], "unverified": [] }, { "id": "alibaba-qwen-image-3-0", "name": "Qwen-Image-3.0", "level": "model", "family_id": "alibaba-qwen-image-family", "developer": [ "alibaba" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-07", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba proprietary image model, Jul 2026.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text" ], "unverified": [] }, { "id": "alibaba-qwen-image-3-0-pro", "name": "Qwen-Image-3.0-Pro", "level": "model", "family_id": "alibaba-qwen-image-family", "developer": [ "alibaba" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-07", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba proprietary image model, Jul 2026.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text", "https://arena.ai/leaderboard/text-to-image" ], "unverified": [] }, { "id": "alibaba-qwen-image-family", "name": "Qwen-Image", "level": "family", "family_id": null, "developer": [ "alibaba" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-08-02", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba Qwen image generation line.", "verified_on": "2026-10-10", "sources": [ "https://datanorth.ai/news/qwen-releases-qwen-image-2-1", "https://huggingface.co/Qwen/Qwen-Image-2.1", "https://huggingface.co/Qwen/Qwen-Image" ], "unverified": [] }, { "id": "alibaba-qwen-image-layered", "name": "Qwen-Image-Layered", "level": "model", "family_id": "alibaba-qwen-image-family", "developer": [ "alibaba" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-12", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen-Image layered model (Dec 2025) merged into Qwen-Image-2.1.", "verified_on": "2026-10-10", "sources": [ "https://datanorth.ai/news/qwen-releases-qwen-image-2-1" ], "unverified": [] }, { "id": "alibaba-qwen-tts-model-studio", "name": "Qwen TTS (Model Studio)", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba Cloud Model Studio TTS; Indonesian not found in the docs text.", "verified_on": "2026-10-10", "sources": [ "https://www.alibabacloud.com/help/en/model-studio/qwen-tts" ], "unverified": [] }, { "id": "alibaba-qwen1-5", "name": "Qwen1.5", "level": "family", "family_id": "alibaba-qwen", "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "Decoder-only transformer with SwiGLU and attention QKV bias; GQA only in 32B and 110B; an MoE 14B A2.7B member also exists", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 32768, "release_date": "2024-02-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "tongyi-qianwen-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Multilingual base and chat models; vendor multilingual evaluation covers 12 languages including Indonesian, Vietnamese and Thai", "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Qwen1.5 is the beta of Qwen2, a family of open decoder-only language models from 0.5B to 110B parameters with 32K context.", "verified_on": "2026-10-11", "sources": [ "https://qwenlm.github.io/blog/qwen1.5/", "https://huggingface.co/Qwen/Qwen1.5-110B-Chat" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "alibaba-qwen1-5-110b-chat", "name": "Qwen1.5-110B-Chat", "level": "model", "family_id": "alibaba-qwen1-5", "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "Transformer decoder with grouped query attention", "params_total": 111209914368, "params_active": null, "params_checkpoint": null, "params_note": "Hugging Face API Safetensors count; named 110B", "context_length": 32000, "release_date": "2024-04-25", "release_date_basis": "announcement page", "weights_date": "2024-04-25", "openness": "open weights restricted", "licence": "tongyi-qianwen-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Blog lists English, Chinese, French, Spanish, German, Russian, Korean, Japanese, Vietnamese, Arabic and others", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Qwen1.5-110B-Chat is the largest dense chat model of the Qwen1.5 series, the first Qwen model over 100B parameters.", "verified_on": "2026-10-11", "sources": [ "https://qwenlm.github.io/blog/qwen1.5-110b/", "https://huggingface.co/Qwen/Qwen1.5-110B-Chat", "https://huggingface.co/api/models/Qwen/Qwen1.5-110B-Chat" ], "unverified": [] }, { "id": "alibaba-qwen1-5-72b", "name": "Qwen1.5-72B", "level": "model", "family_id": "alibaba-qwen1-5", "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 72000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 32768, "release_date": "2024-02-04", "release_date_basis": "announcement page", "weights_date": "2024-01-23", "openness": "open weights restricted", "licence": "tongyi-qianwen-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [ "vllm", "sglang", "llama-cpp", "ollama" ], "languages_note": "Vendor multilingual evaluation covers Arabic, Spanish, French, Portuguese, German, Italian, Russian, Japanese, Korean, Indonesian, Vietnamese and Thai", "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Qwen1.5-72B is the 72B base model of the Qwen1.5 series; its chat version Qwen1.5-72B-Chat was released alongside.", "verified_on": "2026-10-11", "sources": [ "https://qwenlm.github.io/blog/qwen1.5/", "https://huggingface.co/Qwen/Qwen1.5-72B-Chat", "https://huggingface.co/api/models/Qwen/Qwen1.5-72B" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "alibaba-qwen2", "name": "Qwen2", "level": "family", "family_id": null, "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/SeaLLMs" ], "unverified": [] }, { "id": "alibaba-qwen2-5-0-5b", "name": "Qwen2.5 0.5B", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": 500000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-09-15", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Alibaba LLM used as decoder in Grab's 1B vision LLM.", "verified_on": "2026-10-10", "sources": [ "https://engineering.grab.com/custom-vision-llm-at-grab", "https://huggingface.co/Qwen/Qwen2.5-0.5B" ], "unverified": [] }, { "id": "alibaba-qwen2-5-0-5b-instruct", "name": "Qwen2.5-0.5B-Instruct", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-09-16", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen2.5 model named in the genealogy tables of the research notes.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct", "https://huggingface.co/Qwen/Qwen2.5-7B-Instruct", "https://huggingface.co/Qwen/Qwen2.5-0.5B-Instruct" ], "unverified": [] }, { "id": "alibaba-qwen2-5-14b", "name": "Qwen2.5-14B", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-09-15", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen2.5 model named in the genealogy tables of the research notes.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B", "https://huggingface.co/Qwen/Qwen2.5-14B" ], "unverified": [] }, { "id": "alibaba-qwen2-5-32b", "name": "Qwen2.5-32B", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 32000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-09-15", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen2.5 model named in the genealogy tables of the research notes.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B", "https://huggingface.co/baichuan-inc/Baichuan-M2-32B", "https://huggingface.co/Qwen/Qwen2.5-32B" ], "unverified": [] }, { "id": "alibaba-qwen2-5-32b-instruct", "name": "Qwen2.5-32B-Instruct", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": 32000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-09-17", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Open base model fine-tuned by the s1 study.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2501.19393v1", "https://huggingface.co/Qwen/Qwen2.5-32B-Instruct" ], "unverified": [] }, { "id": "alibaba-qwen2-5-72b-instruct", "name": "Qwen2.5-72B-Instruct", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": 72000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-09-16", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Base model of HuatuoGPT-o1-72B.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/FreedomIntelligence/HuatuoGPT-o1-72B", "https://huggingface.co/Qwen/Qwen2.5-72B-Instruct" ], "unverified": [] }, { "id": "alibaba-qwen2-5-7b-instruct", "name": "Qwen2.5-7B-Instruct", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-09-16", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen2.5 model named in the genealogy tables of the research notes.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct", "https://huggingface.co/Qwen/Qwen2.5-7B-Instruct" ], "unverified": [] }, { "id": "alibaba-qwen2-5-coder", "name": "Qwen2.5-Coder", "level": "family", "family_id": null, "developer": [ "alibaba" ], "model_class": "code", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier Qwen coder family (1.5B, 7B, 14B, 32B), still trending on Hugging Face.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/models?search=coder&sort=trending", "https://huggingface.co/SWE-bench/SWE-agent-LM-32B", "https://huggingface.co/OpenHands/openhands-lm-32b-v0.1" ], "unverified": [ "licence" ] }, { "id": "alibaba-qwen2-5-coder-1-5b", "name": "Qwen2.5-Coder 1.5B", "level": "variant", "family_id": "alibaba-qwen2-5-coder", "developer": [ "alibaba" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 1500000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-09-18", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/models?search=coder&sort=trending", "https://huggingface.co/Qwen/Qwen2.5-Coder-1.5B" ], "unverified": [ "licence" ] }, { "id": "alibaba-qwen2-5-coder-14b", "name": "Qwen2.5-Coder 14B", "level": "variant", "family_id": "alibaba-qwen2-5-coder", "developer": [ "alibaba" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 14000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-11-08", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/models?search=coder&sort=trending", "https://huggingface.co/Qwen/Qwen2.5-Coder-14B" ], "unverified": [ "licence" ] }, { "id": "alibaba-qwen2-5-coder-32b-instruct", "name": "Qwen2.5-Coder-32B-Instruct", "level": "variant", "family_id": "alibaba-qwen2-5-coder", "developer": [ "alibaba" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 32000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-11-06", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/SWE-bench/SWE-agent-LM-32B", "https://huggingface.co/OpenHands/openhands-lm-32b-v0.1", "https://huggingface.co/Qwen/Qwen2.5-Coder-32B-Instruct" ], "unverified": [] }, { "id": "alibaba-qwen2-5-coder-7b", "name": "Qwen2.5-Coder 7B", "level": "variant", "family_id": "alibaba-qwen2-5-coder", "developer": [ "alibaba" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 7000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-09-16", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/models?search=coder&sort=trending", "https://huggingface.co/Qwen/Qwen2.5-Coder-7B" ], "unverified": [ "licence" ] }, { "id": "alibaba-qwen2-5-math-1-5b", "name": "Qwen2.5-Math-1.5B", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-09-16", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen2.5 model named in the genealogy tables of the research notes.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B", "https://huggingface.co/Qwen/Qwen2.5-Math-1.5B" ], "unverified": [] }, { "id": "alibaba-qwen2-5-math-7b", "name": "Qwen2.5-Math-7B", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-09-16", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen2.5 model named in the genealogy tables of the research notes.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B", "https://huggingface.co/Qwen/Qwen2.5-Math-7B" ], "unverified": [] }, { "id": "alibaba-qwen2-5-max", "name": "Qwen2.5-Max", "level": "model", "family_id": "alibaba-qwen", "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "Large-scale MoE pre-trained on over 20 trillion tokens; SFT and RLHF", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-01-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "consumer app" ], "deployment_tiers": [], "footprint_note": "Alibaba Cloud API model name qwen-max-2025-01-25; Qwen Chat", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Qwen2.5-Max is a large-scale Qwen mixture-of-experts model served through the Alibaba Cloud API and Qwen Chat, without open weights.", "verified_on": "2026-10-11", "sources": [ "https://qwenlm.github.io/blog/qwen2.5-max/" ], "unverified": [] }, { "id": "alibaba-qwen2-7b", "name": "Qwen2-7B", "level": "model", "family_id": "alibaba-qwen2", "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-06-04", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Language model used as the base of Molmo 7B-D.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/allenai/Molmo-7B-D-0924", "https://huggingface.co/Qwen/Qwen2-7B" ], "unverified": [] }, { "id": "alibaba-qwen2-vl-2b", "name": "Qwen2-VL 2B", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "multimodal LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": 2000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-09-05", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Alibaba vision-language model whose vision encoder Grab uses.", "verified_on": "2026-10-10", "sources": [ "https://engineering.grab.com/custom-vision-llm-at-grab", "https://huggingface.co/Qwen/Qwen2-VL-2B" ], "unverified": [] }, { "id": "alibaba-qwen3", "name": "Qwen3", "level": "family", "family_id": null, "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "dense and MoE", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "0.6B to 235B; small models distilled from flagship knowledge.", "context_length": null, "release_date": "2025-04-29", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "119 languages and dialects (up from 29)", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba's Qwen3 family (base for several derivatives in the notes).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/tokyotech-llm", "https://huggingface.co/Qwen/Qwen3Guard-Stream-4B", "https://www.gladia.io/blog/best-open-source-speech-to-text-models", "https://simonwillison.net/2025/Apr/29/qwen-3/", "https://huggingface.co/api/models/Qwen/Qwen3-0.6B", "https://arxiv.org/abs/2505.09388", "https://qwenlm.github.io/blog/qwen3/", "https://huggingface.co/Qwen/Qwen3-4B" ], "unverified": [] }, { "id": "alibaba-qwen3-0-6b", "name": "Qwen3-0.6B", "level": "model", "family_id": "alibaba-qwen3", "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 751632384, "params_active": null, "params_checkpoint": null, "params_note": "0.6B", "context_length": null, "release_date": "2025-04-27", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server", "mobile or NPU" ], "footprint_note": "Footprint and latency unverified", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "LLM used as base of OmniVoice.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/k2-fsa/OmniVoice", "https://huggingface.co/api/models/Qwen/Qwen3-0.6B", "https://simonwillison.net/2025/Apr/29/qwen-3/", "https://qwenlm.github.io/blog/qwen3/", "https://huggingface.co/Qwen/Qwen3-0.6B" ], "unverified": [ "footprint_note" ] }, { "id": "alibaba-qwen3-0-6b-base", "name": "Qwen3-0.6B-Base", "level": "model", "family_id": "alibaba-qwen3", "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 600000000, "params_active": null, "params_checkpoint": null, "params_note": "0.6B", "context_length": null, "release_date": "2025-04-28", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Base checkpoint for Qwen3-0.6B.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/jinaai/jina-embeddings-v5-text-small", "https://huggingface.co/api/models/Qwen/Qwen3-0.6B", "https://simonwillison.net/2025/Apr/29/qwen-3/", "https://huggingface.co/Qwen/Qwen3-0.6B-Base" ], "unverified": [] }, { "id": "alibaba-qwen3-1-7b", "name": "Qwen3-1.7B", "level": "model", "family_id": "alibaba-qwen3", "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": 1700000000, "params_active": null, "params_checkpoint": null, "params_note": "1.7B", "context_length": null, "release_date": "2025-04-27", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server", "mobile or NPU" ], "footprint_note": "Footprint and latency unverified", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen3 backbone of OneRec-Foundation 1.7B.", "verified_on": "2026-10-10", "sources": [ "https://github.com/Kuaishou-OneRec/OpenOneRec", "https://simonwillison.net/2025/Apr/29/qwen-3/", "https://qwenlm.github.io/blog/qwen3/", "https://huggingface.co/Qwen/Qwen3-0.6B", "https://huggingface.co/Qwen/Qwen3-1.7B" ], "unverified": [ "licence", "footprint_note" ] }, { "id": "alibaba-qwen3-4b", "name": "Qwen3-4B", "level": "model", "family_id": "alibaba-qwen3", "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 4000000000, "params_active": null, "params_checkpoint": null, "params_note": "4B", "context_length": null, "release_date": "2025-04-27", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server", "mobile or NPU" ], "footprint_note": "Footprint and latency unverified", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen3 4B model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Qwen/Qwen3Guard-Stream-4B", "https://huggingface.co/api/models/Qwen/Qwen3-4B", "https://simonwillison.net/2025/Apr/29/qwen-3/", "https://qwenlm.github.io/blog/qwen3/", "https://huggingface.co/Qwen/Qwen3-0.6B", "https://huggingface.co/Qwen/Qwen3-4B" ], "unverified": [ "footprint_note" ] }, { "id": "alibaba-qwen3-4b-base", "name": "Qwen3-4B-Base", "level": "model", "family_id": "alibaba-qwen3", "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "4B", "context_length": null, "release_date": "2025-04-28", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Base checkpoint for Qwen3-4B.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/api/models/Qwen/Qwen3-4B", "https://simonwillison.net/2025/Apr/29/qwen-3/", "https://huggingface.co/Qwen/Qwen3-4B-Base" ], "unverified": [] }, { "id": "alibaba-qwen3-5", "name": "Qwen3.5", "level": "family", "family_id": null, "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Hybrid Gated DeltaNet plus full attention", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 262000, "release_date": "2026-03-02", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba's Qwen3.5 family, Apache 2.0, hybrid DeltaNet architecture; card claims 201 languages.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/collections/Qwen/qwen35", "https://rits.shanghai.nyu.edu/ai/qwen-3-5-small-models-9b-parameters-that-beat-120b" ], "unverified": [] }, { "id": "alibaba-qwen3-5-0-8b", "name": "Qwen3.5-0.8B", "level": "variant", "family_id": "alibaba-qwen3-5", "developer": [ "alibaba" ], "model_class": "multimodal LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Dense; hybrid Gated DeltaNet plus full attention", "params_total": 800000000, "params_active": null, "params_checkpoint": null, "params_note": "0.8B dense", "context_length": 262000, "release_date": "2026-03-02", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU", "offline capable" ], "footprint_note": "About 1.6 GB unquantized (article estimate); 4-bit cuts VRAM about 75%", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen3.5 0.8B small dense model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/collections/Qwen/qwen35", "https://rits.shanghai.nyu.edu/ai/qwen-3-5-small-models-9b-parameters-that-beat-120b", "https://huggingface.co/Qwen/Qwen3.5-0.8B" ], "unverified": [] }, { "id": "alibaba-qwen3-5-122b-a10b", "name": "Qwen3.5-122B-A10B", "level": "model", "family_id": "alibaba-qwen3-5", "developer": [ "alibaba" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Hybrid MoE", "params_total": 122000000000, "params_active": 10000000000, "params_checkpoint": null, "params_note": "122B total, 10B active (per name)", "context_length": 262000, "release_date": "2026-02", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": "Single GPU at FP8 or Int4", "runtimes": [], "languages_note": null, "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba's 122B hybrid MoE model; ranked third on SEA-HELM Indonesian.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/collections/Qwen/qwen35", "https://leaderboard.sea-lion.ai/detailed/SEA", "https://leaderboard.sea-lion.ai/detailed/ID", "https://huggingface.co/Qwen/Qwen3.5-122B-A10B" ], "unverified": [ "indonesia_hosting" ], "indonesian_measured": [ { "benchmark": "SEA-HELM Indonesian leaderboard (ID aggregate, first score column of the Indonesian Competency and Task Scores table)", "score": "77.93 (95% CI -1.29 to +1.25)", "metric": "score 0 to 100, average of 2000 bootstraps with 95% CI", "run_by": "AI Singapore (SEA-HELM leaderboard maintainer)", "url": "https://leaderboard.sea-lion.ai/detailed/ID" } ] }, { "id": "alibaba-qwen3-5-2b", "name": "Qwen3.5-2B", "level": "variant", "family_id": "alibaba-qwen3-5", "developer": [ "alibaba" ], "model_class": "multimodal LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Dense; hybrid Gated DeltaNet plus full attention", "params_total": 2000000000, "params_active": null, "params_checkpoint": null, "params_note": "2B dense", "context_length": 262000, "release_date": "2026-03-02", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU", "laptop" ], "footprint_note": "About 4 GB unquantized (article estimate); 4-bit cuts VRAM about 75%", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen3.5 2B small dense model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/collections/Qwen/qwen35", "https://rits.shanghai.nyu.edu/ai/qwen-3-5-small-models-9b-parameters-that-beat-120b", "https://huggingface.co/Qwen/Qwen3.5-0.8B" ], "unverified": [] }, { "id": "alibaba-qwen3-5-35b-a3b", "name": "Qwen3.5-35B-A3B", "level": "model", "family_id": "alibaba-qwen3-5", "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 35000000000, "params_active": 3000000000, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-03-02", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/collections/Qwen/qwen35" ], "unverified": [] }, { "id": "alibaba-qwen3-5-397b-a17b", "name": "Qwen3.5-397B-A17B", "level": "model", "family_id": "alibaba-qwen3-5", "developer": [ "alibaba" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Gated DeltaNet + Gated Attention, MoE (512 experts, 10 + 1 active), MTP", "params_total": 397000000000, "params_active": 17000000000, "params_checkpoint": null, "params_note": "From model name", "context_length": 262144, "release_date": "2026-02", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "multi-GPU node" ], "footprint_note": "FP8 and GPTQ-Int4 variants", "runtimes": [], "languages_note": "Card claims 201 languages; Indonesian not individually confirmed", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba's 397B MoE model for text, image and video; teacher for SEA-LION v4.5 distillation. Adapters 18, finetunes 41, quants 77.", "verified_on": "2026-10-10", "sources": [ "https://arena.ai/leaderboard/chat/text", "https://huggingface.co/Qwen/Qwen3.5-397B-A17B", "https://huggingface.co/collections/Qwen/qwen35" ], "unverified": [] }, { "id": "alibaba-qwen3-5-4b", "name": "Qwen3.5-4B", "level": "variant", "family_id": "alibaba-qwen3-5", "developer": [ "alibaba" ], "model_class": "multimodal LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Dense; hybrid Gated DeltaNet plus full attention", "params_total": 4000000000, "params_active": null, "params_checkpoint": null, "params_note": "4B dense", "context_length": 262000, "release_date": "2026-03-02", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "laptop", "consumer GPU" ], "footprint_note": "About 8 GB unquantized (article estimate); 4-bit cuts VRAM about 75%", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen3.5 4B small dense model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/collections/Qwen/qwen35", "https://rits.shanghai.nyu.edu/ai/qwen-3-5-small-models-9b-parameters-that-beat-120b", "https://huggingface.co/Qwen/Qwen3.5-0.8B" ], "unverified": [] }, { "id": "alibaba-qwen3-5-9b", "name": "Qwen3.5-9B", "level": "variant", "family_id": "alibaba-qwen3-5", "developer": [ "alibaba" ], "model_class": "multimodal LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Dense; hybrid Gated DeltaNet plus full attention", "params_total": 9000000000, "params_active": null, "params_checkpoint": null, "params_note": "9B dense", "context_length": 262000, "release_date": "2026-03-02", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": "About 18 GB unquantized (article estimate); RTX 3090/4090; 4-bit cuts VRAM about 75%", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen3.5 9B small dense model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/collections/Qwen/qwen35", "https://rits.shanghai.nyu.edu/ai/qwen-3-5-small-models-9b-parameters-that-beat-120b", "https://huggingface.co/Qwen/Qwen3.5-0.8B" ], "unverified": [] }, { "id": "alibaba-qwen3-5-max-preview", "name": "Qwen3.5 Max Preview", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://arena.ai/leaderboard/chat/text" ], "unverified": [] }, { "id": "alibaba-qwen3-5-small", "name": "Qwen3.5 small (0.8B, 2B, 4B, 9B)", "level": "model", "family_id": "alibaba-qwen3-5", "developer": [ "alibaba" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Hybrid DeltaNet", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 262000, "release_date": "2026-03-02", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "laptop", "mobile or NPU", "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": "Likely 201 languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba's small Qwen3.5 sizes; base checkpoints released.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/collections/Qwen/qwen35" ], "unverified": [ "context_length", "modalities_in" ] }, { "id": "alibaba-qwen3-6", "name": "Qwen3.6", "level": "family", "family_id": null, "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04-21", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba's Qwen3.6 family.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Qwen/Qwen3.6-27B" ], "unverified": [] }, { "id": "alibaba-qwen3-6-27b", "name": "Qwen3.6-27B", "level": "model", "family_id": "alibaba-qwen3-6", "developer": [ "alibaba" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Gated DeltaNet + Gated Attention, dense", "params_total": 27000000000, "params_active": null, "params_checkpoint": null, "params_note": "1.01M with YaRN", "context_length": 262144, "release_date": "2026-04", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba's 27B dense model; base of Qwen-SEA-LION v4.5. Adapters 555, finetunes 411, merges 24, quants 743.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Qwen/Qwen3.6-27B", "https://leaderboard.sea-lion.ai/detailed/SEA", "https://leaderboard.sea-lion.ai/detailed/ID", "https://huggingface.co/aisingapore/Qwen-SEA-LION-v4.5-27B-IT" ], "unverified": [ "indonesia_hosting" ], "indonesian_measured": [ { "benchmark": "SEA-HELM Indonesian leaderboard (ID aggregate, first score column of the Indonesian Competency and Task Scores table)", "score": "77.96 (95% CI -1.22 to +1.20)", "metric": "score 0 to 100, average of 2000 bootstraps with 95% CI", "run_by": "AI Singapore (SEA-HELM leaderboard maintainer)", "url": "https://leaderboard.sea-lion.ai/detailed/ID" } ] }, { "id": "alibaba-qwen3-6-35b-a3b", "name": "Qwen3.6-35B-A3B", "level": "model", "family_id": "alibaba-qwen3-6", "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 35000000000, "params_active": 3000000000, "params_checkpoint": null, "params_note": null, "context_length": 262144, "release_date": "2026-04-02", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU", "laptop" ], "footprint_note": "T4/T5 estimate (unverified). About 1M with YaRN.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "General open model marketed as an agentic coder; the strongest small open agentic coder found in the research notes.", "verified_on": "2026-10-10", "sources": [ "https://rits.shanghai.nyu.edu/ai/qwen3-6-35b-a3b-alibaba-open-sources-a-frontier-class-agentic-coder/", "https://huggingface.co/Qwen/Qwen3.6-27B" ], "unverified": [ "deployment_tiers" ] }, { "id": "alibaba-qwen3-6-max-preview", "name": "Qwen3.6 Max Preview", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 1000000000000, "params_active": null, "params_checkpoint": null, "params_note": "About 1T (secondary)", "context_length": 1000000, "release_date": "2026-04", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": "Proprietary Qwen 3.6 Max preview; claimed first on SWE-bench Pro and Terminal-Bench 2.0 without figures (unverified).", "verified_on": "2026-10-10", "sources": [ "https://arena.ai/leaderboard/chat/text", "https://www.bitdoze.com/qwen36-ai-coding-agents/" ], "unverified": [] }, { "id": "alibaba-qwen3-6-plus", "name": "Qwen 3.6 Plus", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": "2026-04", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Proprietary Qwen 3.6 MoE model on Alibaba Cloud API used for coding agents.", "verified_on": "2026-10-10", "sources": [ "https://www.bitdoze.com/qwen36-ai-coding-agents/", "https://leaderboard.steel.dev/leaderboards/swe-bench-verified.md" ], "unverified": [] }, { "id": "alibaba-qwen3-7-max", "name": "Qwen3.7 Max", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba model holding the top MMLU-Pro claim on BenchLM as of 10 Oct 2026.", "verified_on": "2026-10-10", "sources": [ "https://www.benchlm.ai/benchmarks/mmlu-pro" ], "unverified": [] }, { "id": "alibaba-qwen3-7-max-preview", "name": "Qwen3.7 Max Preview", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://arena.ai/leaderboard/chat/text" ], "unverified": [] }, { "id": "alibaba-qwen3-7-plus", "name": "qwen3.7-plus", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://www.alibabacloud.com/help/en/model-studio/models" ], "unverified": [] }, { "id": "alibaba-qwen3-8", "name": "Qwen3.8", "level": "family", "family_id": null, "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-08-08", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba's Qwen3.8 general family; flagship open checkpoint Qwen3.8-2.4T-A95B.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Qwen", "https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B/raw/main/README.md" ], "unverified": [] }, { "id": "alibaba-qwen3-8-2-4t-a95b", "name": "Qwen3.8-2.4T-A95B", "level": "model", "family_id": "alibaba-qwen3-8", "developer": [ "alibaba" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Hybrid Gated DeltaNet + Gated Attention, MoE (512 experts, 10 routed + 1 shared), MTP", "params_total": 2400000000000, "params_active": 95000000000, "params_checkpoint": null, "params_note": "Name indicates 2.4T total, 95B active", "context_length": 262144, "release_date": "2026-08", "release_date_basis": "secondary report", "weights_date": "2026-08-08", "openness": "open weights restricted", "licence": "qwen3-8-max-license", "access_routes": [ "open weights download" ], "deployment_tiers": [ "multi-GPU node" ], "footprint_note": "FP8 variant", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open-weight flagship checkpoint of Qwen3.8; a text-only model that requires thinking mode, under the custom Qwen3.8-Max License.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/api/models/Qwen/Qwen3.8-2.4T-A95B", "https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B/raw/main/README.md", "https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B/raw/main/LICENSE", "https://artificialanalysis.ai/models/qwen3-8-2-4t-a95b", "https://huggingface.co/Qwen", "https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B", "https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B/blob/main/LICENSE", "https://codersera.com/blog/open-source-llms-landscape-2026/", "https://runtimewire.com/article/alibaba-qwen38-24t-open-weights-enterprise-license" ], "unverified": [] }, { "id": "alibaba-qwen3-8-27b", "name": "Qwen3.8-27B", "level": "model", "family_id": "alibaba-qwen3-8", "developer": [ "alibaba" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Gated DeltaNet + Gated Attention, dense, MTP, vision encoder", "params_total": 27000000000, "params_active": null, "params_checkpoint": null, "params_note": "1M with YaRN", "context_length": 262144, "release_date": "2026-08", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": "T4 workstation estimate (derived from parameter count, unverified). 262K native context, 1M extended.", "runtimes": [], "languages_note": "Not yet on SEA-HELM", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Dense 27B open-weight Qwen3.8 model, newest small open agentic coder (Aug 2026).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Qwen/Qwen3.8-27B", "https://huggingface.co/Qwen", "https://arena.ai/leaderboard/chat/text", "https://codersera.com/blog/open-source-llms-landscape-2026/", "https://huggingface.co/api/models/Qwen/Qwen3.8-27B", "https://www.alibabacloud.com/blog/alibaba-unveils-qwen3-8-27b-and-releases-weights-of-qwen3-8-flagship-model_603463" ], "unverified": [ "deployment_tiers" ] }, { "id": "alibaba-qwen3-8-flash", "name": "qwen3.8-flash", "level": "model", "family_id": "alibaba-qwen3-8", "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://www.alibabacloud.com/help/en/model-studio/models" ], "unverified": [] }, { "id": "alibaba-qwen3-8-flash-next", "name": "Qwen3.8-Flash-Next", "level": "model", "family_id": "alibaba-qwen3-8", "developer": [ "alibaba" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Gated DeltaNet + Qwen Sparse Attention, MoE (512 experts, 10 + 1 active), n-gram embedding", "params_total": 125000000000, "params_active": 6000000000, "params_checkpoint": 180000000000, "params_note": "125B core; 180B including 51B n-gram embedding and 4B MTP", "context_length": 262144, "release_date": "2026-08-24", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights restricted", "licence": "qwen-community-1-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": "Single GPU at FP8", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba's mid-size MoE model for image, video and text (1M context with YaRN). Adapters 1, finetunes 71, merges 3, quants 391.", "verified_on": "2026-10-10", "sources": [ "https://codersera.com/blog/open-source-llms-landscape-2026/", "https://huggingface.co/Qwen", "https://huggingface.co/Qwen/Qwen3.8-Flash-Next", "https://huggingface.co/Qwen/Qwen3.8-Flash-Next/blob/main/LICENSE" ], "unverified": [] }, { "id": "alibaba-qwen3-8-max", "name": "Qwen3.8 Max", "level": "model", "family_id": "alibaba-qwen3-8", "developer": [ "alibaba" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "AA lists 984k context (980k in FAQ); model card and Qwen Cloud give 1M by default", "context_length": 1000000, "release_date": "2026-08-03", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Proprietary API version of Qwen3.8 built on the open Qwen3.8-2.4T-A95B checkpoint, adding vision input, non-thinking support and 1M context. No open weights.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/models/qwen3-8-max", "https://www.qwencloud.com/models/qwen3.8-max", "https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B/raw/main/README.md", "https://arena.ai/leaderboard/chat/text", "https://arena.ai/leaderboard/chat/vision", "https://www.benchlm.ai/benchmarks/osworld-verified", "https://datanorth.ai/news/alibaba-releases-qwen3-8-max" ], "unverified": [] }, { "id": "alibaba-qwen3-8-max-prime", "name": "Qwen3.8-Max-Prime", "level": "model", "family_id": "alibaba-qwen3-8", "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Served commercially without weights.", "verified_on": "2026-10-10", "sources": [ "https://codersera.com/blog/open-source-llms-landscape-2026/" ], "unverified": [] }, { "id": "alibaba-qwen3-8-omni-flash", "name": "qwen3.8-omni-flash", "level": "model", "family_id": "alibaba-qwen3-8", "developer": [ "alibaba" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "audio", "video" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Served commercially without weights.", "verified_on": "2026-10-10", "sources": [ "https://www.alibabacloud.com/help/en/model-studio/models", "https://codersera.com/blog/open-source-llms-landscape-2026/" ], "unverified": [] }, { "id": "alibaba-qwen3-8-omni-flash-realtime", "name": "qwen3.8-omni-flash-realtime", "level": "model", "family_id": "alibaba-qwen3-8", "developer": [ "alibaba" ], "model_class": "multimodal LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://www.alibabacloud.com/help/en/model-studio/models" ], "unverified": [] }, { "id": "alibaba-qwen3-8b", "name": "Qwen3-8B", "level": "model", "family_id": "alibaba-qwen3", "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": "8B", "context_length": null, "release_date": "2025-04-27", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen3 backbone of OneRec-Foundation 8B.", "verified_on": "2026-10-10", "sources": [ "https://github.com/Kuaishou-OneRec/OpenOneRec", "https://huggingface.co/MiniMaxAI/MiniMax-Music3", "https://huggingface.co/api/models?filter=base_model:finetune:Qwen/Qwen3-8B", "https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct", "https://huggingface.co/Qwen/Qwen3-8B" ], "unverified": [] }, { "id": "alibaba-qwen3-asr", "name": "Qwen3-ASR", "level": "family", "family_id": null, "developer": [ "alibaba" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "autoregressive", "architecture_detail": "Audio encoder + Qwen3 decoder (unverified detail)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01-29", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU", "CPU server" ], "footprint_note": null, "runtimes": [], "languages_note": "52 languages and dialects per Gladia; technical report: 30 languages and 22 Chinese dialects, Indonesian included", "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba Qwen multilingual ASR family (1.7B and 0.6B open, Flash via API). Most remaining language-ID errors on FLEURS come from confusing Malay and Indonesian.", "verified_on": "2026-10-10", "sources": [ "https://www.gladia.io/blog/best-open-source-speech-to-text-models", "https://benchmarklist.com/benchmarks/open_asr_leaderboard/", "https://arxiv.org/html/2601.21337v2" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "alibaba-qwen3-asr-0-6b", "name": "Qwen3-ASR-0.6B", "level": "variant", "family_id": "alibaba-qwen3-asr", "developer": [ "alibaba" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "autoregressive", "architecture_detail": null, "params_total": 600000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01-29", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU", "CPU server" ], "footprint_note": "CPU-feasible (judgement)", "runtimes": [], "languages_note": null, "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen3-ASR 0.6B variant.", "verified_on": "2026-10-10", "sources": [ "https://www.gladia.io/blog/best-open-source-speech-to-text-models", "https://benchmarklist.com/benchmarks/open_asr_leaderboard/", "https://arxiv.org/html/2601.21337v2" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "alibaba-qwen3-asr-1-7b", "name": "Qwen3-ASR-1.7B", "level": "variant", "family_id": "alibaba-qwen3-asr", "developer": [ "alibaba" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "autoregressive", "architecture_detail": null, "params_total": 1700000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01-29", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen3-ASR 1.7B variant.", "verified_on": "2026-10-10", "sources": [ "https://www.gladia.io/blog/best-open-source-speech-to-text-models", "https://benchmarklist.com/benchmarks/open_asr_leaderboard/", "https://arxiv.org/html/2601.21337v2" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "alibaba-qwen3-asr-flash-1208", "name": "Qwen3-ASR-Flash-1208", "level": "variant", "family_id": "alibaba-qwen3-asr", "developer": [ "alibaba" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "API reference model in the Qwen3-ASR technical report.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2601.21337v2" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "alibaba-qwen3-coder", "name": "Qwen3-Coder", "level": "family", "family_id": null, "developer": [ "alibaba" ], "model_class": "code", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-07-22", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Qwen/Qwen3-Coder-Next", "https://huggingface.co/Qwen/Qwen3-Coder-480B-A35B-Instruct", "https://huggingface.co/Qwen/Qwen3-Coder-30B-A3B-Instruct" ], "unverified": [] }, { "id": "alibaba-qwen3-coder-30b-a3b-instruct", "name": "Qwen3-Coder-30B-A3B-Instruct", "level": "model", "family_id": "alibaba-qwen3-coder", "developer": [ "alibaba" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 30500000000, "params_active": 3300000000, "params_checkpoint": null, "params_note": null, "context_length": 262144, "release_date": "2025", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU", "laptop" ], "footprint_note": "T4/T5 estimate (unverified). 1M with YaRN.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small Qwen3 open coder model; no scores on card.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Qwen/Qwen3-Coder-30B-A3B-Instruct" ], "unverified": [ "deployment_tiers" ] }, { "id": "alibaba-qwen3-coder-480b-a35b-instruct", "name": "Qwen3-Coder-480B-A35B-Instruct", "level": "model", "family_id": "alibaba-qwen3-coder", "developer": [ "alibaba" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 480000000000, "params_active": 35000000000, "params_checkpoint": null, "params_note": null, "context_length": 262144, "release_date": "2025", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "multi-GPU node" ], "footprint_note": "T3 single 8-GPU server estimate (unverified). 1M with YaRN.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Large Qwen3 open coder model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Qwen/Qwen3-Coder-480B-A35B-Instruct" ], "unverified": [ "deployment_tiers" ] }, { "id": "alibaba-qwen3-coder-next", "name": "Qwen3-Coder-Next", "level": "model", "family_id": "alibaba-qwen3-coder", "developer": [ "alibaba" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Gated DeltaNet plus gated attention hybrid MoE, 512 experts; non-thinking only", "params_total": 80000000000, "params_active": 3000000000, "params_checkpoint": null, "params_note": null, "context_length": 262144, "release_date": "2026-02-03", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": "T4 estimate; needs memory for 80B weights (unverified).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba's open coder model, 80B total and 3B active; offered free in Kiro.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Qwen/Qwen3-Coder-Next", "https://kiro.dev/docs/models/" ], "unverified": [ "deployment_tiers" ] }, { "id": "alibaba-qwen3-embedding-0-6b", "name": "Qwen3-Embedding-0.6B", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 600000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 32000, "release_date": "2025-06", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "100+ languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba Qwen3-Embedding-0.6B on a Qwen3 base. Embedding dimension up to 4096 (32 to 4096 adjustable).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Qwen/Qwen3-Embedding-8B", "https://huggingface.co/Qwen/models?search=embedding", "https://huggingface.co/Qwen/Qwen3-Reranker-8B" ], "unverified": [] }, { "id": "alibaba-qwen3-embedding-4b", "name": "Qwen3-Embedding-4B", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 4000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 32000, "release_date": "2025-06", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "100+ languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba Qwen3-Embedding-4B on a Qwen3 base. Embedding dimension up to 4096 (32 to 4096 adjustable).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Qwen/Qwen3-Embedding-8B", "https://huggingface.co/Qwen/models?search=embedding", "https://huggingface.co/Qwen/Qwen3-Reranker-8B" ], "unverified": [] }, { "id": "alibaba-qwen3-embedding-8b", "name": "Qwen3-Embedding-8B", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 32000, "release_date": "2025-06", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "100+ languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba Qwen3-Embedding-8B on a Qwen3 base. Embedding dimension up to 4096 (32 to 4096 adjustable).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Qwen/Qwen3-Embedding-8B", "https://huggingface.co/Qwen/models?search=embedding", "https://huggingface.co/Qwen/Qwen3-Reranker-8B" ], "unverified": [] }, { "id": "alibaba-qwen3-reranker-0-6b", "name": "Qwen3-Reranker-0.6B", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "reranker", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 600000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 32000, "release_date": "2025-06", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "100+ languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba Qwen3-Reranker-0.6B on a Qwen3 base.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Qwen/Qwen3-Embedding-8B", "https://huggingface.co/Qwen/models?search=embedding", "https://huggingface.co/Qwen/Qwen3-Reranker-8B" ], "unverified": [] }, { "id": "alibaba-qwen3-reranker-4b", "name": "Qwen3-Reranker-4B", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "reranker", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 4000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 32000, "release_date": "2025-06", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "100+ languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba Qwen3-Reranker-4B on a Qwen3 base.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Qwen/Qwen3-Embedding-8B", "https://huggingface.co/Qwen/models?search=embedding", "https://huggingface.co/Qwen/Qwen3-Reranker-8B" ], "unverified": [] }, { "id": "alibaba-qwen3-reranker-8b", "name": "Qwen3-Reranker-8B", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "reranker", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 32000, "release_date": "2025-06", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "100+ languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba Qwen3-Reranker-8B on a Qwen3 base.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Qwen/Qwen3-Embedding-8B", "https://huggingface.co/Qwen/models?search=embedding", "https://huggingface.co/Qwen/Qwen3-Reranker-8B" ], "unverified": [] }, { "id": "alibaba-qwen3-tts-12hz-0-6b", "name": "Qwen3-TTS 12Hz 0.6B", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "autoregressive", "architecture_detail": null, "params_total": 600000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01-22", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Smaller open Qwen TTS model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Qwen/Qwen3-TTS-12Hz-1.7B-CustomVoice" ], "unverified": [] }, { "id": "alibaba-qwen3-tts-12hz-1-7b", "name": "Qwen3-TTS 12Hz 1.7B", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "autoregressive", "architecture_detail": "Discrete multi-codebook LM", "params_total": 1700000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01-22", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": "10 languages; Indonesian not listed", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open Qwen TTS model (also 0.6B size).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Qwen/Qwen3-TTS-12Hz-1.7B-CustomVoice" ], "unverified": [] }, { "id": "alibaba-qwen3-vl-8b-instruct", "name": "Qwen3-VL-8B-Instruct", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10-11", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/aisingapore/Qwen-SEA-Guard-8B-2602", "https://huggingface.co/Qwen/Qwen3-VL-8B-Instruct" ], "unverified": [] }, { "id": "alibaba-qwen3-vl-embedding-2b", "name": "Qwen3-VL-Embedding-2B", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 2000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 32000, "release_date": "2026-04", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "100+ languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba Qwen3-VL-Embedding-2B on a Qwen3 base. Embedding dimension up to 4096 (32 to 4096 adjustable).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Qwen/Qwen3-Embedding-8B", "https://huggingface.co/Qwen/models?search=embedding", "https://huggingface.co/Qwen/Qwen3-Reranker-8B" ], "unverified": [] }, { "id": "alibaba-qwen3-vl-embedding-8b", "name": "Qwen3-VL-Embedding-8B", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 32000, "release_date": "2026-04", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "100+ languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba Qwen3-VL-Embedding-8B on a Qwen3 base. Embedding dimension up to 4096 (32 to 4096 adjustable).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Qwen/Qwen3-Embedding-8B", "https://huggingface.co/Qwen/models?search=embedding", "https://huggingface.co/Qwen/Qwen3-Reranker-8B" ], "unverified": [] }, { "id": "alibaba-qwen3guard", "name": "Qwen3Guard", "level": "family", "family_id": null, "developer": [ "alibaba" ], "model_class": "guard", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10-16", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "119 languages (Indonesian not individually confirmed)", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba's guard models with three severity tiers (Safe, Controversial, Unsafe) across categories including violence, illegal acts, sexual, PII, self-harm, unethical acts, politically sensitive, copyright and jailbreak (input only). Stream variants moderate token by token.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Qwen/Qwen3Guard-Stream-4B" ], "unverified": [] }, { "id": "alibaba-qwen3guard-gen-0-6b", "name": "Qwen3Guard-Gen-0.6B", "level": "model", "family_id": "alibaba-qwen3guard", "developer": [ "alibaba" ], "model_class": "guard", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 600000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10-16", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "119 languages (Indonesian not individually confirmed)", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen3Guard gen guard model. Gen card not read (rate limit).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Qwen/Qwen3Guard-Stream-4B" ], "unverified": [ "summary" ] }, { "id": "alibaba-qwen3guard-gen-4b", "name": "Qwen3Guard-Gen-4B", "level": "model", "family_id": "alibaba-qwen3guard", "developer": [ "alibaba" ], "model_class": "guard", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 4000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10-16", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "119 languages (Indonesian not individually confirmed)", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen3Guard gen guard model. Gen card not read (rate limit).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Qwen/Qwen3Guard-Stream-4B" ], "unverified": [ "summary" ] }, { "id": "alibaba-qwen3guard-gen-8b", "name": "Qwen3Guard-Gen-8B", "level": "model", "family_id": "alibaba-qwen3guard", "developer": [ "alibaba" ], "model_class": "guard", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10-16", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "119 languages (Indonesian not individually confirmed)", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen3Guard gen guard model. Gen card not read (rate limit).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Qwen/Qwen3Guard-Stream-4B" ], "unverified": [ "summary" ] }, { "id": "alibaba-qwen3guard-stream-0-6b", "name": "Qwen3Guard-Stream-0.6B", "level": "model", "family_id": "alibaba-qwen3guard", "developer": [ "alibaba" ], "model_class": "guard", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 600000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10-16", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "119 languages (Indonesian not individually confirmed)", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen3Guard stream guard model. Stream repos refreshed about Sep 2026.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Qwen/Qwen3Guard-Stream-4B" ], "unverified": [] }, { "id": "alibaba-qwen3guard-stream-4b", "name": "Qwen3Guard-Stream-4B", "level": "model", "family_id": "alibaba-qwen3guard", "developer": [ "alibaba" ], "model_class": "guard", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 4000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10-16", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "119 languages (Indonesian not individually confirmed)", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen3Guard stream guard model. Stream repos refreshed about Sep 2026.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Qwen/Qwen3Guard-Stream-4B" ], "unverified": [] }, { "id": "alibaba-qwen3guard-stream-8b", "name": "Qwen3Guard-Stream-8B", "level": "model", "family_id": "alibaba-qwen3guard", "developer": [ "alibaba" ], "model_class": "guard", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10-16", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "119 languages (Indonesian not individually confirmed)", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen3Guard stream guard model. Stream repos refreshed about Sep 2026.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Qwen/Qwen3Guard-Stream-4B" ], "unverified": [] }, { "id": "alibaba-qwq-32b", "name": "QwQ-32B", "level": "model", "family_id": "alibaba-qwen", "developer": [ "alibaba" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 32500000000, "params_active": null, "params_checkpoint": null, "params_note": "31.0B non-embedding parameters; 64 layers", "context_length": 131072, "release_date": "2025-03-06", "release_date_basis": "announcement page", "weights_date": "2025-03-06", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "consumer app" ], "deployment_tiers": [], "footprint_note": "YaRN must be enabled for prompts over 8,192 tokens", "runtimes": [], "languages_note": "Hugging Face language tag: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen 32B reasoning model trained with reinforcement learning, reported by Qwen as comparable to DeepSeek-R1.", "verified_on": "2026-10-11", "sources": [ "https://qwenlm.github.io/blog/qwq-32b/", "https://huggingface.co/Qwen/QwQ-32B" ], "unverified": [] }, { "id": "alibaba-qwq-32b-preview", "name": "QwQ-32B-Preview", "level": "model", "family_id": "alibaba-qwen", "developer": [ "alibaba" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "Causal transformer with RoPE, SwiGLU, RMSNorm and attention QKV bias; 64 layers; GQA 40 Q and 8 KV heads", "params_total": 32500000000, "params_active": null, "params_checkpoint": null, "params_note": "31.0B non-embedding", "context_length": 32768, "release_date": "2024-11-28", "release_date_basis": "announcement page", "weights_date": "2024-11-27", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Card language tag: English; card warns of language mixing and code-switching", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": "QwQ-32B-Preview is an experimental Qwen research model focused on reasoning, strong in math and coding, fine-tuned from Qwen2.5-32B-Instruct.", "verified_on": "2026-10-11", "sources": [ "https://qwenlm.github.io/blog/qwq-32b-preview/", "https://huggingface.co/Qwen/QwQ-32B-Preview", "https://huggingface.co/api/models/Qwen/QwQ-32B-Preview" ], "unverified": [] }, { "id": "alibaba-sensevoicesmall", "name": "SenseVoiceSmall", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "Non-autoregressive encoder (unverified)", "params_total": 234000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-07-03", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "funasr-model-license", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server", "embedded GPU" ], "footprint_note": null, "runtimes": [], "languages_note": "zh/en/ja/ko/yue", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ASR plus emotion and audio-event model; HF licence tag 'other' (name 'model-license').", "verified_on": "2026-10-10", "sources": [ "https://github.com/modelscope/FunASR", "https://huggingface.co/api/models/FunAudioLLM/SenseVoiceSmall" ], "unverified": [] }, { "id": "alibaba-wan", "name": "Wan", "level": "family", "family_id": null, "developer": [ "alibaba" ], "model_class": "video generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba Tongyi Wan video model line.", "verified_on": "2026-10-10", "sources": [ "https://www.atlascloud.ai/de/blog/tips/is-wan-3.0-open-source" ], "unverified": [] }, { "id": "alibaba-wan-2-1", "name": "Wan 2.1", "level": "model", "family_id": "alibaba-wan", "developer": [ "alibaba" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "1.3B and 14B sizes", "context_length": null, "release_date": "2025", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Early-2025 open Wan video model, Apache 2.0.", "verified_on": "2026-10-10", "sources": [ "https://www.atlascloud.ai/de/blog/tips/is-wan-3.0-open-source", "https://huggingface.co/spaces/Vchitect/VBench_Leaderboard" ], "unverified": [] }, { "id": "alibaba-wan-2-2", "name": "Wan 2.2", "level": "family", "family_id": "alibaba-wan", "developer": [ "alibaba" ], "model_class": "video generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-07-28", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Last open Wan flagship generation, Apache 2.0.", "verified_on": "2026-10-10", "sources": [ "https://www.atlascloud.ai/de/blog/tips/is-wan-3.0-open-source", "https://huggingface.co/models?pipeline_tag=text-to-video&sort=trending" ], "unverified": [] }, { "id": "alibaba-wan-2-2-a14b", "name": "Wan 2.2 A14B", "level": "model", "family_id": "alibaba-wan-2-2", "developer": [ "alibaba" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": "MoE", "architecture_detail": null, "params_total": 27000000000, "params_active": 14000000000, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-07-28", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Last open Wan flagship, MoE with 27B total and 14B active parameters.", "verified_on": "2026-10-10", "sources": [ "https://www.atlascloud.ai/de/blog/tips/is-wan-3.0-open-source", "https://huggingface.co/models?pipeline_tag=text-to-video&sort=trending" ], "unverified": [] }, { "id": "alibaba-wan-2-2-ti2v-5b", "name": "Wan 2.2 TI2V-5B", "level": "model", "family_id": "alibaba-wan-2-2", "developer": [ "alibaba" ], "model_class": "video generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": 5000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-07-28", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "5B text-and-image-to-video Wan 2.2 model.", "verified_on": "2026-10-10", "sources": [ "https://www.atlascloud.ai/de/blog/tips/is-wan-3.0-open-source", "https://huggingface.co/models?pipeline_tag=text-to-video&sort=trending" ], "unverified": [ "deployment_tiers" ] }, { "id": "alibaba-wan-2-5", "name": "Wan 2.5", "level": "model", "family_id": "alibaba-wan", "developer": [ "alibaba" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "API-only Wan model.", "verified_on": "2026-10-10", "sources": [ "https://www.atlascloud.ai/de/blog/tips/is-wan-3.0-open-source" ], "unverified": [] }, { "id": "alibaba-wan-2-6", "name": "Wan 2.6", "level": "model", "family_id": "alibaba-wan", "developer": [ "alibaba" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "API-only Wan model.", "verified_on": "2026-10-10", "sources": [ "https://www.atlascloud.ai/de/blog/tips/is-wan-3.0-open-source" ], "unverified": [] }, { "id": "alibaba-wan-2-7", "name": "Wan 2.7", "level": "model", "family_id": "alibaba-wan", "developer": [ "alibaba" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "API-only Wan model.", "verified_on": "2026-10-10", "sources": [ "https://www.atlascloud.ai/de/blog/tips/is-wan-3.0-open-source", "https://arena.ai/leaderboard/text-to-video" ], "unverified": [] }, { "id": "alibaba-wan-3-0", "name": "Wan 3.0", "level": "model", "family_id": "alibaba-wan", "developer": [ "alibaba" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-08-06", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "Cloud only; API beta", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": "Alibaba's proprietary video model, API beta from 6 Aug 2026, no weights; 480p to 1080p, up to 30 s per a third-party listing. Leads the Artificial Analysis text-to-video board.", "verified_on": "2026-10-10", "sources": [ "https://www.atlascloud.ai/de/blog/tips/is-wan-3.0-open-source", "https://artificialanalysis.ai/video/leaderboard/text-to-video", "https://arena.ai/leaderboard/text-to-video", "https://en.wikipedia.org/wiki/Alibaba_Group" ], "unverified": [] }, { "id": "alibaba-wan-dancer-14b", "name": "Wan-Dancer-14B", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "avatar", "modalities_in": [ "image", "video" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": 14000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-07", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open Wan character animation control model.", "verified_on": "2026-10-10", "sources": [ "https://www.atlascloud.ai/de/blog/tips/is-wan-3.0-open-source" ], "unverified": [] }, { "id": "alibaba-wan2-2-animate-2-14b", "name": "Wan2.2-Animate-2-14B", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "avatar", "modalities_in": [ "image", "video" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": 14000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-08", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open Wan character animation control model.", "verified_on": "2026-10-10", "sources": [ "https://www.atlascloud.ai/de/blog/tips/is-wan-3.0-open-source" ], "unverified": [] }, { "id": "alibaba-z-image-base", "name": "Z-Image Base", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Base model of Z-Image (Tongyi-MAI).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Tongyi-MAI/Z-Image-Turbo" ], "unverified": [] }, { "id": "alibaba-z-image-turbo", "name": "Z-Image Turbo", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": "Single-Stream DiT (S3-DiT); developer team Tongyi-MAI", "params_total": 6000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-11-27", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": "Fits 16 GB", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba Tongyi-MAI 6B open image model, Apache 2.0, 623k Hugging Face downloads.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Tongyi-MAI/Z-Image-Turbo", "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text", "https://huggingface.co/models?pipeline_tag=text-to-image&sort=trending", "https://en.wikipedia.org/wiki/Alibaba_Group" ], "unverified": [] }, { "id": "all-hands-ai-openhands-lm-32b-v0-1", "name": "OpenHands LM 32B v0.1", "level": "model", "family_id": null, "developer": [ "all-hands-ai" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 32000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03-31", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Research agent model fine-tuned from Qwen2.5-Coder-32B-Instruct.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/OpenHands/openhands-lm-32b-v0.1" ], "unverified": [] }, { "id": "all-hands-ai-swe-gym-agents", "name": "SWE-Gym agents", "level": "model", "family_id": null, "developer": [ "all-hands-ai" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-12", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "SWE-Gym paper listed on the All Hands HF org; no public models in Oct 2026.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/all-hands" ], "unverified": [ "status" ] }, { "id": "allen-ai-molmo", "name": "Molmo", "level": "family", "family_id": null, "developer": [ "allen-ai" ], "model_class": "multimodal LLM", "modalities_in": [ "image", "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-09-25", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Molmo is a family of open vision-language models from the Allen Institute for AI, trained on the PixMo image-text dataset.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/allenai/Molmo-7B-D-0924" ], "unverified": [] }, { "id": "allen-ai-molmo-2", "name": "Molmo 2", "level": "family", "family_id": null, "developer": [ "allen-ai" ], "model_class": "multimodal LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-12-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Molmo 2 is the Ai2 family of open vision-language models for image, multi-image and video understanding and grounding.", "verified_on": "2026-10-11", "sources": [ "https://allenai.org/blog/molmo2" ], "unverified": [] }, { "id": "allen-ai-molmo-7b-d", "name": "Molmo 7B-D", "level": "model", "family_id": "allen-ai-molmo", "developer": [ "allen-ai" ], "model_class": "multimodal LLM", "modalities_in": [ "image", "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "Qwen2-7B language model with an OpenAI CLIP (ViT-L/14 336) vision backbone", "params_total": 8021025280, "params_active": null, "params_checkpoint": null, "params_note": "Safetensors parameter count from the Hugging Face API", "context_length": null, "release_date": "2024-09-25", "release_date_basis": "registry date", "weights_date": "2024-09-25", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Card language tag: English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": "Molmo 7B-D is an open vision-language model from the Allen Institute for AI built on Qwen2-7B and CLIP, intended for research and educational use.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/allenai/Molmo-7B-D-0924", "https://huggingface.co/api/models/allenai/Molmo-7B-D-0924" ], "unverified": [] }, { "id": "allen-ai-molmo2-8b", "name": "Molmo2-8B", "level": "model", "family_id": "allen-ai-molmo-2", "developer": [ "allen-ai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "Qwen3-8B language model with SigLIP 2 vision encoder", "params_total": 8660000000, "params_active": null, "params_checkpoint": null, "params_note": "Named 8B; Hugging Face lists 9B; Artificial Analysis gives 8.66B", "context_length": null, "release_date": "2025-12-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Ai2 vision-language model for image, multi-image and video understanding with pointing, tracking and grounded video QA.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/allenai/Molmo2-8B", "https://allenai.org/blog/molmo2", "https://artificialanalysis.ai/models/molmo2-8b" ], "unverified": [] }, { "id": "allen-ai-olmo-2", "name": "OLMo 2", "level": "family", "family_id": null, "developer": [ "allen-ai" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-11", "release_date_basis": "inferred from page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "OLMo 2 is the Ai2 family of fully open language models at 7B, 13B and 32B.", "verified_on": "2026-10-11", "sources": [ "https://allenai.org/blog/olmo2-32B" ], "unverified": [] }, { "id": "allen-ai-olmo-2-0325-32b", "name": "OLMo-2-0325-32B", "level": "model", "family_id": "allen-ai-olmo-2", "developer": [ "allen-ai" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03-12", "release_date_basis": "registry date", "weights_date": null, "openness": "fully open", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Base OLMo 2 32B model pretrained on up to 6T tokens of OLMo-Mix-1124 plus the Dolmino mid-training mix.", "verified_on": "2026-10-11", "sources": [ "https://allenai.org/blog/olmo2-32B", "https://huggingface.co/allenai/OLMo-2-0325-32B-Instruct" ], "unverified": [] }, { "id": "allen-ai-olmo-2-32b", "name": "OLMo 2 32B (OLMo-2-0325-32B-Instruct)", "level": "model", "family_id": "allen-ai-olmo-2", "developer": [ "allen-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 32000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "fully open", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Primarily English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Ai2 fully open 32B model with released data, code and weights, which Ai2 says outperforms GPT-3.5 Turbo and GPT-4o mini on academic benchmarks.", "verified_on": "2026-10-11", "sources": [ "https://allenai.org/blog/olmo2-32B", "https://huggingface.co/allenai/OLMo-2-0325-32B-Instruct" ], "unverified": [] }, { "id": "allen-ai-olmo-3", "name": "Olmo 3", "level": "family", "family_id": null, "developer": [ "allen-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-11-19", "release_date_basis": "registry date", "weights_date": null, "openness": "fully open", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Allen AI's fully open Olmo 3 family: Dolma 3 pretraining data, Dolci post-training data, OLMo-Core and Open-Instruct code, all checkpoints. No Olmo 4 found.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/allenai/Olmo-3-32B-Think", "https://huggingface.co/allenai/models?search=olmo" ], "unverified": [] }, { "id": "allen-ai-olmo-3-1-32b-instruct", "name": "Olmo 3.1 32B Instruct", "level": "model", "family_id": "allen-ai-olmo-3", "developer": [ "allen-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Transformer-style autoregressive language model", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "32B per model name; Artificial Analysis lists 32.2B total", "context_length": 66000, "release_date": "2025-12-10", "release_date_basis": "registry date", "weights_date": "2025-12-10", "openness": "fully open", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": "Licensed Apache 2.0 for research and educational use under Ai2 Responsible Use Guidelines; data cutoff Dec 2024", "runtimes": [], "languages_note": "English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Ai2 fully open 32B instruction-tuned model for chat, tool use and multi-turn dialogue, post-trained from Olmo 3 32B with SFT, DPO and RLVR.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/allenai/Olmo-3.1-32B-Instruct", "https://allenai.org/blog/olmo3", "https://artificialanalysis.ai/models/olmo-3-1-32b-instruct" ], "unverified": [] }, { "id": "allen-ai-olmo-3-1-32b-think", "name": "Olmo 3.1 32B Think", "level": "model", "family_id": "allen-ai-olmo-3", "developer": [ "allen-ai" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 32000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "fully open", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Newer 3.1 version of Olmo 3 32B Think.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/allenai/Olmo-3-32B-Think", "https://huggingface.co/allenai/models?search=olmo" ], "unverified": [] }, { "id": "allen-ai-olmo-3-1125-32b", "name": "Olmo-3-1125-32B", "level": "model", "family_id": "allen-ai-olmo-3", "developer": [ "allen-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 32000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "fully open", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/allenai/Olmo-3-32B-Think" ], "unverified": [] }, { "id": "allen-ai-olmo-3-32b-think", "name": "Olmo 3 32B Think", "level": "model", "family_id": "allen-ai-olmo-3", "developer": [ "allen-ai" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 32000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-12-15", "release_date_basis": "notes", "weights_date": null, "openness": "fully open", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU", "consumer GPU" ], "footprint_note": "Consumer GPU at 4-bit", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Allen AI's fully open 32B reasoning model. Finetunes 6, quants 33.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/allenai/Olmo-3-32B-Think", "https://huggingface.co/allenai/models?search=olmo" ], "unverified": [] }, { "id": "allen-ai-olmo-3-7b", "name": "Olmo 3 7B", "level": "model", "family_id": "allen-ai-olmo-3", "developer": [ "allen-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 7000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025", "release_date_basis": "notes", "weights_date": null, "openness": "fully open", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Allen AI's fully open 7B model; many RL-Zero variants.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/allenai/models?search=olmo" ], "unverified": [] }, { "id": "allen-ai-olmo-hybrid-7b", "name": "Olmo-Hybrid-7B", "level": "model", "family_id": "allen-ai-olmo-3", "developer": [ "allen-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": null, "params_total": 7000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-05", "release_date_basis": "registry date", "weights_date": null, "openness": "fully open", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Allen AI's hybrid-architecture 7B model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/allenai/models?search=olmo" ], "unverified": [] }, { "id": "allen-ai-wildguard-7b", "name": "WildGuard 7B", "level": "model", "family_id": null, "developer": [ "allen-ai" ], "model_class": "guard", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 7000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-06", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Allen AI's guard model for prompt harm, response harm and refusal detection; paper reports beating Llama-Guard2 and Aegis-Guard, up to +25.3% F1 on refusal detection.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/allenai/wildguard" ], "unverified": [] }, { "id": "allen-institute-for-ai-olmocr", "name": "olmOCR (olmOCR-2-7B-1025)", "level": "model", "family_id": null, "developer": [ "allen-ai" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "autoregressive", "architecture_detail": null, "params_total": 8290000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10-06", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Ai2 OCR VLM with its own olmOCR-Bench; not in the OmniDocBench v1.6 table.", "verified_on": "2026-10-10", "sources": [ "https://benchmarklist.com/benchmarks/olmocr_bench/", "https://huggingface.co/api/models/allenai/olmOCR-2-7B-1025" ], "unverified": [] }, { "id": "amazon-autogluon-1-5", "name": "AutoGluon 1.5", "level": "model", "family_id": null, "developer": [ "amazon" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": "AutoML ensemble (includes GBDTs and TFMs)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026", "release_date_basis": "notes", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server", "single datacentre GPU" ], "footprint_note": "4h budget in TabArena; median train about 289 s; latest PyPI autogluon 1.6.3 on 18 Sep 2026", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Amazon AutoML ensemble; rank 1 on TabArena (extreme, 4h).", "verified_on": "2026-10-10", "sources": [ "https://benchmarklist.com/benchmarks/tabarena_all/", "https://pypi.org/pypi/autogluon/json", "https://github.com/autogluon/autogluon" ], "unverified": [] }, { "id": "amazon-chronos", "name": "Chronos", "level": "model", "family_id": null, "developer": [ "amazon" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "T5-based, 20M to 710M parameters; public plus synthetic Gaussian-process data.", "context_length": null, "release_date": "2024-03", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First Chronos forecasting model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/amazon/chronos-2", "https://www.amazon.science/blog/introducing-chronos-2-from-univariate-to-universal-forecasting", "https://arxiv.org/abs/2403.07815" ], "unverified": [] }, { "id": "amazon-chronos-2", "name": "Chronos-2", "level": "model", "family_id": null, "developer": [ "amazon" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "dense transformer", "architecture_detail": "Encoder-only, T5-encoder-inspired; group attention for cross-series and covariate in-context learning; quantile output; horizon 1,024", "params_total": 120000000, "params_active": null, "params_checkpoint": null, "params_note": "about 120M", "context_length": 8192, "release_date": "2025-10-20", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "single datacentre GPU", "CPU server" ], "footprint_note": "More than 300 forecasts/s on one A10G", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Amazon (AWS AI) zero-shot forecasting model with covariate in-context learning; report 17 Oct 2025. Context is 8,192 time steps.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/amazon/chronos-2", "https://www.beri.net/article/timesfm-vs-chronos-vs-moirai-time-series-foundation-model-licence-bigquery-sagemaker-cost-2026", "https://arxiv.org/pdf/2605.20119", "https://www.amazon.science/blog/introducing-chronos-2-from-univariate-to-universal-forecasting" ], "unverified": [] }, { "id": "amazon-chronos-bolt", "name": "Chronos-Bolt", "level": "model", "family_id": null, "developer": [ "amazon" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Chronos variant preceding Chronos-2.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/amazon/chronos-2", "https://www.amazon.science/blog/introducing-chronos-2-from-univariate-to-universal-forecasting" ], "unverified": [] }, { "id": "amazon-mitra", "name": "Mitra", "level": "model", "family_id": null, "developer": [ "amazon" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": "Synthetic priors + fine-tuning", "params_total": 75700000, "params_active": null, "params_checkpoint": null, "params_note": "mitra-classifier", "context_length": null, "release_date": "2025", "release_date_basis": "notes", "weights_date": "2025-06-22", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Amazon AWS tabular foundation model.", "verified_on": "2026-10-10", "sources": [ "https://mindfulmodeler.substack.com/p/the-state-of-tabular-foundation-models", "https://www.beri.net/article/mitra-vs-tabpfn-vs-xgboost-tabular-foundation-model-licence-benchmarks-2026", "https://huggingface.co/api/models/autogluon/mitra-classifier" ], "unverified": [] }, { "id": "amazon-nova", "name": "Nova", "level": "family", "family_id": null, "developer": [ "amazon" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://aws.amazon.com/nova/" ], "unverified": [] }, { "id": "amazon-nova-2-0-omni", "name": "Nova 2.0 Omni", "level": "model", "family_id": "amazon-nova", "developer": [ "amazon" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/leaderboards/models" ], "unverified": [] }, { "id": "amazon-nova-2-0-pro-preview", "name": "Nova 2.0 Pro Preview", "level": "model", "family_id": "amazon-nova", "developer": [ "amazon" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 256000, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/leaderboards/models" ], "unverified": [] }, { "id": "amazon-nova-2-lite", "name": "Nova 2 Lite", "level": "model", "family_id": "amazon-nova", "developer": [ "amazon" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "cross-region only", "status": "active", "summary": "Amazon reasoning model; on Bedrock Jakarta via Global cross-Region only.", "verified_on": "2026-10-10", "sources": [ "https://aws.amazon.com/nova/", "https://docs.aws.amazon.com/bedrock/latest/userguide/models-region-compatibility.html" ], "unverified": [] }, { "id": "amazon-nova-2-sonic", "name": "Nova 2 Sonic", "level": "model", "family_id": "amazon-nova", "developer": [ "amazon" ], "model_class": "speech synthesis", "modalities_in": [ "audio" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Amazon speech-to-speech model.", "verified_on": "2026-10-10", "sources": [ "https://aws.amazon.com/nova/" ], "unverified": [] }, { "id": "amazon-nova-act", "name": "Nova Act", "level": "model", "family_id": "amazon-nova", "developer": [ "amazon" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Browser-agent service built on a custom Nova 2 Lite.", "verified_on": "2026-10-10", "sources": [ "https://aws.amazon.com/nova/" ], "unverified": [] }, { "id": "amazon-nova-forge", "name": "Nova Forge", "level": "model", "family_id": "amazon-nova", "developer": [ "amazon" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Nova offering listed on aws.amazon.com/nova; nature not described in the notes.", "verified_on": "2026-10-10", "sources": [ "https://aws.amazon.com/nova/" ], "unverified": [ "model_class" ] }, { "id": "amazon-nova-lite", "name": "Nova Lite", "level": "model", "family_id": "amazon-nova", "developer": [ "amazon" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "in-region Jakarta", "status": "active", "summary": "Nova model hosted in-Region in Bedrock Jakarta (ap-southeast-3) and via Geo cross-Region.", "verified_on": "2026-10-10", "sources": [ "https://docs.aws.amazon.com/bedrock/latest/userguide/models-region-compatibility.html" ], "unverified": [] }, { "id": "amazon-nova-micro", "name": "Nova Micro", "level": "model", "family_id": "amazon-nova", "developer": [ "amazon" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Nova model available in Singapore via Geo cross-Region.", "verified_on": "2026-10-10", "sources": [ "https://docs.aws.amazon.com/bedrock/latest/userguide/models-region-compatibility.html" ], "unverified": [] }, { "id": "amazon-nova-premier", "name": "Nova Premier", "level": "model", "family_id": "amazon-nova", "developer": [ "amazon" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "End of life 2026-09-14 (secondary report).", "verified_on": "2026-10-10", "sources": [ "https://slash-digital.io/en/insights/ai-models-2026/" ], "unverified": [] }, { "id": "amazon-nova-pro", "name": "Nova Pro", "level": "model", "family_id": "amazon-nova", "developer": [ "amazon" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "in-region Jakarta", "status": "active", "summary": "Nova model hosted in-Region in Bedrock Jakarta (ap-southeast-3) and via Geo cross-Region.", "verified_on": "2026-10-10", "sources": [ "https://docs.aws.amazon.com/bedrock/latest/userguide/models-region-compatibility.html" ], "unverified": [] }, { "id": "ampixa-sanotts", "name": "sanoTTS", "level": "model", "family_id": null, "developer": [ "ampixa" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": "Distilled tiny neural TTS voices: piperlite lineage from a Piper/VITS teacher; nano lineage (mel-100, ConvNeXt1D, iSTFT) from a Kokoro teacher", "params_total": 2272145, "params_active": null, "params_checkpoint": null, "params_note": "Voices range from 294,279 to 2,272,145 parameters; the ESP32-S3 int8 voice has 567,008", "context_length": null, "release_date": "2026-07-13", "release_date_basis": "registry date", "weights_date": "2026-07-13", "openness": "open source", "licence": "gpl-3-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "microcontroller", "laptop", "CPU server", "offline capable" ], "footprint_note": "337 KB to 8.7 MB per voice; real time on an ESP32-S3; runs in the browser via WebAssembly; pure numpy inference", "runtimes": [], "languages_note": "11 voices in English, Nepali, Hindi, Vietnamese, Indonesian and Chinese", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "sanoTTS is a family of tiny distilled text-to-speech voices that run without cloud or NPU on a USD 3 ESP32-S3 microcontroller or in the browser, including an Indonesian voice.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/ampixa/sanoTTS", "https://huggingface.co/api/models/ampixa/sanoTTS" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "ant-group-skysense", "name": "SkySense", "level": "model", "family_id": null, "developer": [ "ant-group", "wuhan-university", "mybank" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": "Factorized multi-modal spatiotemporal encoder (optical + SAR time series); Multi-Granularity Contrastive Learning; Geo-Context Prototype Learning; pre-trained on 21.5M temporal sequences", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "billion-scale", "context_length": null, "release_date": "2023-12-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "hyperscale cluster", "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Ant Group remote sensing foundation model (CVPR 2024); 16 datasets, 7 tasks; claims to beat 18 RSFMs.", "verified_on": "2026-10-10", "sources": [ "https://openaccess.thecvf.com/content/CVPR2024/html/Guo_SkySense_A_Multi-Modal_Remote_Sensing_Foundation_Model_Towards_Universal_Interpretation_CVPR_2024_paper.html", "https://arxiv.org/html/2312.10115v2" ], "unverified": [ "licence" ] }, { "id": "ant-group-skysense-o", "name": "SkySense-O", "level": "model", "family_id": null, "developer": [ "ant-group" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "SkySense variant named in the research notes; not fetched.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2312.10115v2" ], "unverified": [ "developer" ] }, { "id": "ant-group-skysense-v2", "name": "SkySense V2", "level": "model", "family_id": null, "developer": [ "ant-group" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "MoE", "architecture_detail": "Single unified transformer backbone for multiple modalities; adaptive patch merging; modality prompt tokens; MoE", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-07-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Successor of SkySense (ICCV 2025); beats SkySense by 1.8 points on average over 16 datasets.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2507.13812" ], "unverified": [ "developer" ] }, { "id": "ant-group-xihe-ultra", "name": "Xihe-ultra", "level": "model", "family_id": null, "developer": [ "ant-group" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Ant time-series model on GIFT-Eval.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2605.20119", "https://huggingface.co/spaces/Salesforce/GIFT-Eval" ], "unverified": [] }, { "id": "anthropic-claude", "name": "Claude", "level": "family", "family_id": null, "developer": [ "anthropic" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Anthropic's proprietary model family, including the Opus, Sonnet, Haiku, Fable and Mythos tiers; no open-weight relatives.", "verified_on": "2026-10-10", "sources": [ "https://www.anthropic.com/news", "https://platform.claude.com/docs/en/models/overview" ], "unverified": [] }, { "id": "anthropic-claude-3-5-haiku", "name": "Claude 3.5 Haiku", "level": "model", "family_id": null, "developer": [ "anthropic" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Production Anthropic model studied with a cross-layer transcoder of about 30M features in On the Biology of a Large Language Model.", "verified_on": "2026-10-10", "sources": [ "https://transformer-circuits.pub/2025/attribution-graphs/biology.html" ], "unverified": [] }, { "id": "anthropic-claude-3-5-sonnet-oct-24", "name": "Claude 3.5 Sonnet (Oct '24)", "level": "model", "family_id": "anthropic-claude", "developer": [ "anthropic" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-10-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "consumer app", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": "Computer use beta on the Anthropic API, Amazon Bedrock and Google Cloud Vertex AI", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Upgraded Claude 3.5 Sonnet released with the public beta of computer use, with large gains on agentic coding and tool-use benchmarks over the earlier Claude 3.5 Sonnet.", "verified_on": "2026-10-11", "sources": [ "https://www.anthropic.com/news/3-5-models-and-computer-use" ], "unverified": [] }, { "id": "anthropic-claude-3-7-sonnet", "name": "Claude 3.7 Sonnet", "level": "model", "family_id": "anthropic-claude", "developer": [ "anthropic" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier Claude model used by SWE-agent to generate the SWE-smith trajectories for SWE-agent-LM-32B.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/SWE-bench/SWE-agent-LM-32B" ], "unverified": [] }, { "id": "anthropic-claude-3-haiku", "name": "Claude 3 Haiku", "level": "model", "family_id": "anthropic-claude", "developer": [ "anthropic" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-03-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "consumer app", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": "Claude API, claude.ai for Pro subscribers, Amazon Bedrock; Vertex AI announced as coming soon", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Claude 3 Haiku is the fastest and lowest-cost model of the Claude 3 family at launch, with vision input, aimed at enterprise workloads.", "verified_on": "2026-10-11", "sources": [ "https://www.anthropic.com/news/claude-3-haiku" ], "unverified": [] }, { "id": "anthropic-claude-3-opus", "name": "Claude 3 Opus", "level": "model", "family_id": "anthropic-claude", "developer": [ "anthropic" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 200000, "release_date": "2024-03-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "consumer app" ], "deployment_tiers": [], "footprint_note": "200K context at launch; inputs above 1 million tokens offered to select customers; API id claude-3-opus-20240229 retired 5 January 2026", "runtimes": [], "languages_note": "Announcement cites improved fluency in non-English languages such as Spanish, Japanese and French", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "shut down", "summary": "Most capable model of the Claude 3 family, aimed at highly complex tasks with vision input. Anthropic retired it on the Claude API on 5 January 2026.", "verified_on": "2026-10-11", "sources": [ "https://www.anthropic.com/news/claude-3-family", "https://www-cdn.anthropic.com/de8ba9b01c9ab7cbabf5c33b80b7bbc618857627/Model_Card_Claude_3.pdf", "https://platform.claude.com/docs/en/about-claude/model-deprecations.md" ], "unverified": [] }, { "id": "anthropic-claude-3-sonnet", "name": "Claude 3 Sonnet", "level": "model", "family_id": "anthropic-claude", "developer": [ "anthropic" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 200000, "release_date": "2024-03-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "consumer app", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": "API id claude-3-sonnet-20240229 retired 21 July 2025", "runtimes": [], "languages_note": "Announcement cites improved fluency in non-English languages such as Spanish, Japanese and French", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "shut down", "summary": "Claude 3 family model balancing intelligence and speed for enterprise workloads, with vision input. Anthropic retired it on the Claude API on 21 July 2025.", "verified_on": "2026-10-11", "sources": [ "https://www.anthropic.com/news/claude-3-family", "https://www-cdn.anthropic.com/de8ba9b01c9ab7cbabf5c33b80b7bbc618857627/Model_Card_Claude_3.pdf", "https://platform.claude.com/docs/en/about-claude/model-deprecations.md" ], "unverified": [] }, { "id": "anthropic-claude-fable-5", "name": "Claude Fable 5", "level": "model", "family_id": "anthropic-claude", "developer": [ "anthropic" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed; shares architecture with restricted Claude Mythos (secondary report)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 2000000, "release_date": "2026-06-09", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Public Claude 5 release derived from the Mythos line; withdrawn 2026-06-12 and returned 2026-07-01. GA in GitHub Copilot.", "verified_on": "2026-10-10", "sources": [ "https://www.anthropic.com/news", "https://slash-digital.io/en/insights/ai-models-2026/", "https://vpsranking.com/news/ai/ai-2026-09-22-anthropic-claude-opus-5-5/", "https://arena.ai/leaderboard/chat/text", "https://leaderboard.steel.dev/leaderboards/swe-bench-verified.md", "https://ecosistemastartup.com/claude-fable-5-93-9-en-swe-bench-y-lo-que-significa-para-founders/", "https://llm-stats.com/blog/research/claude-opus-5-5-launch" ], "unverified": [ "context_length" ] }, { "id": "anthropic-claude-fable-5-1", "name": "Claude Fable 5.1", "level": "model", "family_id": "anthropic-claude", "developer": [ "anthropic" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": "2026-09-01", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "Vendor cloud/API only (T1). Max output 128K tokens.", "runtimes": [], "languages_note": "Anthropic docs state all current models support multilingual capabilities; Bahasa Indonesia not verified", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Anthropic's model for demanding reasoning and long-horizon agentic work; adaptive thinking always on, default effort high. Available via API, GitHub Copilot (admin opt-in) and Kiro limited preview.", "verified_on": "2026-10-10", "sources": [ "https://platform.claude.com/docs/en/models/overview", "https://slash-digital.io/en/insights/ai-models-2026/", "https://artificialanalysis.ai/leaderboards/models", "https://vals.ai/models/anthropic_claude-fable-5-1", "https://platform.claude.com/docs/en/about-claude/models/overview", "https://docs.github.com/en/copilot/reference/ai-models/supported-models", "https://kiro.dev/docs/models/", "https://www.benchlm.ai/benchmarks/hle", "https://benchlm.ai/coding", "https://www.benchlm.ai/benchmarks/mmmu" ], "unverified": [ "release_date" ] }, { "id": "anthropic-claude-haiku-4-5", "name": "Claude Haiku 4.5", "level": "model", "family_id": "anthropic-claude", "developer": [ "anthropic" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Predecessor of Claude Haiku 5.5, supported until 2026-10-15.", "verified_on": "2026-10-10", "sources": [ "https://www.anthropic.com/news", "https://slash-digital.io/en/insights/ai-models-2026/", "https://vpsranking.com/news/ai/ai-2026-09-22-anthropic-claude-opus-5-5/", "https://arena.ai/leaderboard/chat/text" ], "unverified": [] }, { "id": "anthropic-claude-haiku-5-5", "name": "Claude Haiku 5.5", "level": "model", "family_id": "anthropic-claude", "developer": [ "anthropic" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": "2026-10-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": "Vendor cloud/API only (T1). Max output 128K. Bedrock Jakarta (ap-southeast-3) and Singapore via Global cross-Region inference only.", "runtimes": [], "languages_note": "Anthropic docs state all current models support multilingual capabilities; Bahasa Indonesia not verified", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "cross-region only", "status": "active", "summary": "Anthropic's model for high-volume, cost-sensitive work; adaptive thinking, default effort medium.", "verified_on": "2026-10-10", "sources": [ "https://platform.claude.com/docs/en/models/overview", "https://www.anthropic.com/news", "https://docs.aws.amazon.com/bedrock/latest/userguide/models-region-compatibility.html", "https://platform.claude.com/docs/en/about-claude/models/overview" ], "unverified": [] }, { "id": "anthropic-claude-mythos-5", "name": "Claude Mythos 5", "level": "model", "family_id": "anthropic-claude", "developer": [ "anthropic" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "gated preview" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Restricted-access (dual-use) tier referenced via the Cyber Verification Program; no public benchmarks in the main model notes (the coding notes cite a SWE-bench Verified score).", "verified_on": "2026-10-10", "sources": [ "https://www.anthropic.com/news", "https://slash-digital.io/en/insights/ai-models-2026/", "https://vpsranking.com/news/ai/ai-2026-09-22-anthropic-claude-opus-5-5/", "https://arena.ai/leaderboard/chat/text", "https://leaderboard.steel.dev/leaderboards/swe-bench-verified.md", "https://ecosistemastartup.com/claude-fable-5-93-9-en-swe-bench-y-lo-que-significa-para-founders/", "https://llm-stats.com/blog/research/claude-opus-5-5-launch" ], "unverified": [ "access_routes" ] }, { "id": "anthropic-claude-mythos-5-1", "name": "Claude Mythos 5.1", "level": "model", "family_id": "anthropic-claude", "developer": [ "anthropic" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "gated preview" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Restricted-access tier referenced via the Cyber Verification Program.", "verified_on": "2026-10-10", "sources": [ "https://www.anthropic.com/news", "https://slash-digital.io/en/insights/ai-models-2026/", "https://vpsranking.com/news/ai/ai-2026-09-22-anthropic-claude-opus-5-5/", "https://arena.ai/leaderboard/chat/text" ], "unverified": [ "access_routes" ] }, { "id": "anthropic-claude-opus-4", "name": "Claude Opus 4", "level": "model", "family_id": "anthropic-claude", "developer": [ "anthropic" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": "Hybrid model with near-instant responses and an extended thinking mode", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-05-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": "API model id claude-opus-4-20250514; retired on the Claude API on 15 June 2026", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "shut down", "summary": "Anthropic Claude 4 generation model aimed at coding, long-running agent workflows and advanced reasoning. Anthropic retired it on the Claude API on 15 June 2026.", "verified_on": "2026-10-11", "sources": [ "https://www.anthropic.com/news/claude-4", "https://platform.claude.com/docs/en/about-claude/model-deprecations.md" ], "unverified": [] }, { "id": "anthropic-claude-opus-4-1", "name": "Claude Opus 4.1", "level": "model", "family_id": "anthropic-claude", "developer": [ "anthropic" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-08-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "consumer app", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": "API model id claude-opus-4-1-20250805; retired on the Claude API on 5 August 2026 (partner platforms set their own dates)", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "shut down", "summary": "Anthropic upgrade to Claude Opus 4 for coding, agentic tasks and reasoning, offered in Claude apps, Claude Code, the API, Amazon Bedrock and Vertex AI. Anthropic retired it on the Claude API on 5 August 2026.", "verified_on": "2026-10-11", "sources": [ "https://www.anthropic.com/news/claude-opus-4-1", "https://platform.claude.com/docs/en/about-claude/model-deprecations.md" ], "unverified": [] }, { "id": "anthropic-claude-opus-4-5", "name": "Claude Opus 4.5", "level": "model", "family_id": null, "developer": [ "anthropic" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Anthropic model behind Claude for Healthcare (11 Jan 2026), HIPAA-ready, with connectors for CMS Coverage, ICD-10, NPI, PubMed, Medidata and ClinicalTrials.gov.", "verified_on": "2026-10-10", "sources": [ "https://www.anthropic.com/news/healthcare-life-sciences" ], "unverified": [] }, { "id": "anthropic-claude-opus-4-6", "name": "Claude Opus 4.6", "level": "model", "family_id": "anthropic-claude", "developer": [ "anthropic" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-02", "release_date_basis": "notes", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Claude Opus model from Feb 2026, still on Arena (Text #3, Vision #3).", "verified_on": "2026-10-10", "sources": [ "https://www.anthropic.com/news", "https://slash-digital.io/en/insights/ai-models-2026/", "https://vpsranking.com/news/ai/ai-2026-09-22-anthropic-claude-opus-5-5/", "https://arena.ai/leaderboard/chat/text" ], "unverified": [] }, { "id": "anthropic-claude-opus-4-7", "name": "Claude Opus 4.7", "level": "model", "family_id": "anthropic-claude", "developer": [ "anthropic" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier Claude Opus model; retired from GitHub Copilot 2026-10-02.", "verified_on": "2026-10-10", "sources": [ "https://docs.github.com/en/copilot/reference/ai-models/supported-models" ], "unverified": [] }, { "id": "anthropic-claude-opus-4-8", "name": "Claude Opus 4.8", "level": "model", "family_id": "anthropic-claude", "developer": [ "anthropic" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Claude Opus model still listed on Arena; GA in GitHub Copilot.", "verified_on": "2026-10-10", "sources": [ "https://www.anthropic.com/news", "https://slash-digital.io/en/insights/ai-models-2026/", "https://vpsranking.com/news/ai/ai-2026-09-22-anthropic-claude-opus-5-5/", "https://arena.ai/leaderboard/chat/text" ], "unverified": [] }, { "id": "anthropic-claude-opus-5", "name": "Claude Opus 5", "level": "model", "family_id": "anthropic-claude", "developer": [ "anthropic" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Predecessor of Claude Opus 5.5, still available; release date not stated.", "verified_on": "2026-10-10", "sources": [ "https://www.anthropic.com/news", "https://slash-digital.io/en/insights/ai-models-2026/", "https://vpsranking.com/news/ai/ai-2026-09-22-anthropic-claude-opus-5-5/", "https://arena.ai/leaderboard/chat/text", "https://leaderboard.steel.dev/leaderboards/swe-bench-verified.md", "https://ecosistemastartup.com/claude-fable-5-93-9-en-swe-bench-y-lo-que-significa-para-founders/", "https://llm-stats.com/blog/research/claude-opus-5-5-launch", "https://benchlm.ai/benchmarks/swe-bench-verified" ], "unverified": [] }, { "id": "anthropic-claude-opus-5-5", "name": "Claude Opus 5.5", "level": "model", "family_id": "anthropic-claude", "developer": [ "anthropic" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": "2026-09-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": "Vendor cloud/API only (T1). Max output 128K (300K on Batch API). API ID claude-opus-5-5; also on Amazon Bedrock, Google Cloud Vertex AI, Microsoft Foundry, Claude Platform on AWS.", "runtimes": [], "languages_note": "Anthropic docs state all current models support multilingual capabilities; Bahasa Indonesia not verified", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Anthropic's recommended default model, stated to perform at the level of Claude Fable 5.1 on most work. Used for coding via Claude Code, GitHub Copilot and Kiro.", "verified_on": "2026-10-10", "sources": [ "https://platform.claude.com/docs/en/models/opus-5-5/overview", "https://platform.claude.com/docs/en/models/overview", "https://www.anthropic.com/news", "https://artificialanalysis.ai/models/claude-opus-5-5", "https://llm-stats.com/blog/research/claude-opus-5-5-launch", "https://platform.claude.com/docs/en/about-claude/models/overview", "https://www.benchlm.ai/benchmarks/swePro", "https://epoch.ai/benchmarks", "https://arena.ai/leaderboard" ], "unverified": [] }, { "id": "anthropic-claude-sonnet-4", "name": "Claude 4 Sonnet", "level": "model", "family_id": null, "developer": [ "anthropic" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "cross-region only", "status": null, "summary": "Included in Cognite's Atlas AI LLM benchmark report.", "verified_on": "2026-10-10", "sources": [ "https://cognite.com/en/atlas-ai-slm-llm-benchmark-report", "https://docs.aws.amazon.com/bedrock/latest/userguide/models-region-compatibility.md" ], "unverified": [ "indonesian_measured" ], "hosting_provider": "Amazon Bedrock (ap-southeast-3 Jakarta)", "hosting_url": "https://docs.aws.amazon.com/bedrock/latest/userguide/models-region-compatibility.md" }, { "id": "anthropic-claude-sonnet-4-5", "name": "Claude Sonnet 4.5", "level": "model", "family_id": null, "developer": [ "anthropic" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "cross-region only", "status": null, "summary": "Anthropic model behind Claude for Life Sciences (20 Oct 2025); scores 0.83 on Protocol QA against a 0.79 human baseline. Also benchmarked by Cognite for Atlas AI.", "verified_on": "2026-10-10", "sources": [ "https://www.anthropic.com/news/claude-for-life-sciences", "https://cognite.com/en/atlas-ai-slm-llm-benchmark-report", "https://docs.aws.amazon.com/bedrock/latest/userguide/models-region-compatibility.md" ], "unverified": [ "indonesian_measured" ], "hosting_provider": "Amazon Bedrock (ap-southeast-3 Jakarta)", "hosting_url": "https://docs.aws.amazon.com/bedrock/latest/userguide/models-region-compatibility.md" }, { "id": "anthropic-claude-sonnet-4-6", "name": "Claude Sonnet 4.6", "level": "model", "family_id": "anthropic-claude", "developer": [ "anthropic" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Claude Sonnet model listed as GA in GitHub Copilot.", "verified_on": "2026-10-10", "sources": [ "https://docs.github.com/en/copilot/reference/ai-models/supported-models" ], "unverified": [] }, { "id": "anthropic-claude-sonnet-5", "name": "Claude Sonnet 5", "level": "model", "family_id": "anthropic-claude", "developer": [ "anthropic" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Predecessor of Claude Sonnet 5.5.", "verified_on": "2026-10-10", "sources": [ "https://www.anthropic.com/news", "https://slash-digital.io/en/insights/ai-models-2026/", "https://vpsranking.com/news/ai/ai-2026-09-22-anthropic-claude-opus-5-5/", "https://arena.ai/leaderboard/chat/text" ], "unverified": [] }, { "id": "anthropic-claude-sonnet-5-5", "name": "Claude Sonnet 5.5", "level": "model", "family_id": "anthropic-claude", "developer": [ "anthropic" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": "2026-09-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "Vendor cloud/API only (T1). Max output 128K. Jakarta availability on Bedrock unverified.", "runtimes": [], "languages_note": "Anthropic docs state all current models support multilingual capabilities; Bahasa Indonesia not verified", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Upgrade over Claude Sonnet 5, faster and cheaper; adaptive thinking, default effort high.", "verified_on": "2026-10-10", "sources": [ "https://platform.claude.com/docs/en/models/overview", "https://www.anthropic.com/news", "https://platform.claude.com/docs/en/about-claude/models/overview" ], "unverified": [] }, { "id": "anysphere-composer", "name": "Composer", "level": "family", "family_id": null, "developer": [ "anysphere" ], "model_class": "code", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://framia.converge.ai/page/en-US/news/cursor-composer-2-5-kimi-coding-model", "https://cursor.com/blog/composer-2" ], "unverified": [] }, { "id": "anysphere-composer-2", "name": "Composer 2", "level": "model", "family_id": "anysphere-composer", "developer": [ "anysphere" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Continued pretraining of an unnamed base, then RL", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-03-19", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Cursor coding model.", "verified_on": "2026-10-10", "sources": [ "https://cursor.com/blog/composer-2" ], "unverified": [] }, { "id": "anysphere-composer-2-5", "name": "Composer 2.5", "level": "model", "family_id": "anysphere-composer", "developer": [ "anysphere" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-05-19", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app" ], "deployment_tiers": [], "footprint_note": "Cursor only. Partly trained on Colossus 2.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Cursor's coding model, post-trained from Kimi K2.5 with 25x more synthetic coding tasks than Composer 2; parity claims with Opus 4.7 and GPT-5.5 unscored (unverified).", "verified_on": "2026-10-10", "sources": [ "https://framia.converge.ai/page/en-US/news/cursor-composer-2-5-kimi-coding-model" ], "unverified": [] }, { "id": "apple-on-device-foundation-model-2025", "name": "Apple on-device foundation model (2025)", "level": "model", "family_id": null, "developer": [ "apple" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "2-bit quantization-aware training; KV-cache sharing", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "About 3B", "context_length": null, "release_date": "2025-07-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "internal only" ], "deployment_tiers": [ "mobile or NPU", "laptop", "offline capable" ], "footprint_note": "Footprint not stated; Apple silicon (iPhone, iPad, Mac)", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Apple's about 3B on-device model, accessed only through the Swift Foundation Models framework (guided generation, tool calling, LoRA adapters).", "verified_on": "2026-10-10", "sources": [ "https://machinelearning.apple.com/research/apple-foundation-models-tech-report-2025", "https://arxiv.org/abs/2507.13575v2", "https://arxiv.org/abs/2507.13575" ], "unverified": [] }, { "id": "apple-server-pt-moe-model-2025", "name": "Apple server PT-MoE model (Private Cloud Compute)", "level": "model", "family_id": null, "developer": [ "apple" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "PT-MoE", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "internal only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Apple server model on Private Cloud Compute, paired with the on-device model.", "verified_on": "2026-10-10", "sources": [ "https://machinelearning.apple.com/research/apple-foundation-models-tech-report-2025" ], "unverified": [] }, { "id": "arc-institute-evo-2", "name": "Evo 2", "level": "family", "family_id": null, "developer": [ "arc-institute", "nvidia" ], "model_class": "scientific", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Context up to 1M base pairs", "context_length": 1000000, "release_date": "2025-02-13", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "DNA language model family trained on OpenGenome2 (8.8T tokens), sizes 1B to 40B, Apache-2.0, published in Nature; hosted on NVIDIA BioNeMo and NIM.", "verified_on": "2026-10-10", "sources": [ "https://github.com/ArcInstitute/evo2" ], "unverified": [] }, { "id": "arc-institute-evo-2-1b", "name": "Evo 2 1B", "level": "model", "family_id": "arc-institute-evo-2", "developer": [ "arc-institute", "nvidia" ], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": null, "architecture_detail": null, "params_total": 1000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": "2025-02-13", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Evo 2 DNA language model at 1B parameters.", "verified_on": "2026-10-10", "sources": [ "https://github.com/ArcInstitute/evo2" ], "unverified": [] }, { "id": "arc-institute-evo-2-20b", "name": "Evo 2 20B", "level": "model", "family_id": "arc-institute-evo-2", "developer": [ "arc-institute", "nvidia" ], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": null, "architecture_detail": null, "params_total": 20000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": "2025-02-13", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Evo 2 DNA language model at 20B parameters.", "verified_on": "2026-10-10", "sources": [ "https://github.com/ArcInstitute/evo2" ], "unverified": [] }, { "id": "arc-institute-evo-2-40b", "name": "Evo 2 40B", "level": "model", "family_id": "arc-institute-evo-2", "developer": [ "arc-institute", "nvidia" ], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": null, "architecture_detail": null, "params_total": 40000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": "2025-02-13", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Evo 2 DNA language model at 40B parameters.", "verified_on": "2026-10-10", "sources": [ "https://github.com/ArcInstitute/evo2" ], "unverified": [] }, { "id": "arc-institute-evo-2-7b", "name": "Evo 2 7B", "level": "model", "family_id": "arc-institute-evo-2", "developer": [ "arc-institute", "nvidia" ], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": null, "architecture_detail": null, "params_total": 7000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": "2025-02-13", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Evo 2 DNA language model at 7B parameters.", "verified_on": "2026-10-10", "sources": [ "https://github.com/ArcInstitute/evo2" ], "unverified": [] }, { "id": "arcee-ai-trinity-large", "name": "Trinity-Large", "level": "family", "family_id": null, "developer": [ "arcee-ai" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Arcee AI Trinity-Large family of sparse MoE models (TrueBase, Base, Preview and Thinking).", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/arcee-ai/Trinity-Large-Thinking" ], "unverified": [] }, { "id": "arcee-ai-trinity-large-base", "name": "Trinity-Large-Base", "level": "model", "family_id": "arcee-ai-trinity-large", "developer": [ "arcee-ai" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Arcee AI 17T-token pretrained foundation model with mid-training anneals, base of Trinity-Large-Thinking.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/arcee-ai/Trinity-Large-Thinking" ], "unverified": [] }, { "id": "arcee-ai-trinity-large-thinking", "name": "Trinity-Large-Thinking", "level": "model", "family_id": "arcee-ai-trinity-large", "developer": [ "arcee-ai" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "Sparse MoE with 4 active experts; explicit reasoning traces in think blocks", "params_total": 398000000000, "params_active": 13000000000, "params_checkpoint": null, "params_note": "About 398B total and about 13B active per token", "context_length": 512000, "release_date": "2026-04-01", "release_date_basis": "announcement page", "weights_date": "2026-04-01", "openness": "open weights permissive", "licence": "openmdw-1-1", "access_routes": [ "API", "open weights download" ], "deployment_tiers": [], "footprint_note": "Blog says Apache 2.0, but the LICENSE file in the weights repository is OpenMDW-1.1; the licence file governs", "runtimes": [ "vllm" ], "languages_note": "Hugging Face language tags: en, es, fr, de, it, pt, ru, ar, hi, ko, zh", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Arcee AI open reasoning model for long-horizon agents and multi-turn tool calling, post-trained from Trinity-Large-Base.", "verified_on": "2026-10-11", "sources": [ "https://www.arcee.ai/blog/trinity-large-thinking", "https://huggingface.co/arcee-ai/Trinity-Large-Thinking", "https://huggingface.co/arcee-ai/Trinity-Large-Thinking/raw/main/LICENSE", "https://openrouter.ai/arcee-ai/trinity-large-thinking" ], "unverified": [] }, { "id": "archesweathergen", "name": "ArchesWeatherGen", "level": "model", "family_id": null, "developer": [ "inria" ], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-12-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Ensemble weather model on WeatherBench 2 (2020).", "verified_on": "2026-10-10", "sources": [ "https://researchsites.withgoogle.com/weatherbench/deterministic/", "https://arxiv.org/html/2412.12971", "https://arxiv.org/abs/2412.12971" ], "unverified": [] }, { "id": "artime", "name": "ARTime", "level": "model", "family_id": null, "developer": [], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Top of the NAB v1.0 scoreboard (standard profile).", "verified_on": "2026-10-10", "sources": [ "https://github.com/numenta/NAB" ], "unverified": [] }, { "id": "asymflux-2-klein", "name": "AsymFLUX.2 klein", "level": "model", "family_id": null, "developer": [ "stanford-university" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "flow matching", "architecture_detail": "Pixel-space asymmetric flow model finetuned from FLUX.2 klein Base 9B; base frozen, input and output projections and rank-256 LoRA adapters trained", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Built on the 9B FLUX.2 klein Base; no separate total given", "context_length": null, "release_date": "2026-05", "release_date_basis": "inferred from page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "AsymFLUX.2 klein, a pixel-space text-to-image model finetuned from FLUX.2 klein Base 9B with the AsymFlow method.", "verified_on": "2026-10-11", "sources": [ "https://arxiv.org/abs/2605.12964" ], "unverified": [] }, { "id": "auto-arima", "name": "Auto_Arima", "level": "model", "family_id": null, "developer": [], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Statistical baseline on GIFT-Eval.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/spaces/Salesforce/GIFT-Eval" ], "unverified": [] }, { "id": "baai-bge-m3", "name": "bge-m3", "level": "model", "family_id": null, "developer": [ "baai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BAAI bge-m3 (BGE line; bases XLM-R, Gemma 2, Qwen3; 0.1B to 9B). MIT or Apache, per-model licence unverified.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/BAAI/models?search=bge" ], "unverified": [ "licence" ] }, { "id": "baai-bge-multilingual-gemma2", "name": "bge-multilingual-gemma2", "level": "model", "family_id": null, "developer": [ "baai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BAAI bge-multilingual-gemma2 (BGE line; bases XLM-R, Gemma 2, Qwen3; 0.1B to 9B). MIT or Apache, per-model licence unverified.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/BAAI/models?search=bge" ], "unverified": [ "licence" ] }, { "id": "baai-bge-reasoner-embed-qwen3-8b", "name": "bge-reasoner-embed-qwen3-8b", "level": "model", "family_id": null, "developer": [ "baai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BAAI bge-reasoner-embed-qwen3-8b (BGE line; bases XLM-R, Gemma 2, Qwen3; 0.1B to 9B). MIT or Apache, per-model licence unverified.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/BAAI/models?search=bge" ], "unverified": [ "licence" ] }, { "id": "baai-bge-reranker-v2-m3", "name": "bge-reranker-v2-m3", "level": "model", "family_id": null, "developer": [ "baai" ], "model_class": "reranker", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BAAI bge-reranker-v2-m3 (BGE line; bases XLM-R, Gemma 2, Qwen3; 0.1B to 9B). MIT or Apache, per-model licence unverified.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/BAAI/models?search=bge" ], "unverified": [ "licence" ] }, { "id": "baai-bge-vl", "name": "BGE-VL family", "level": "family", "family_id": null, "developer": [ "baai" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BAAI BGE-VL family (BGE line; bases XLM-R, Gemma 2, Qwen3; 0.1B to 9B). MIT or Apache, per-model licence unverified.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/BAAI/models?search=bge" ], "unverified": [ "licence" ] }, { "id": "bagel-7b-mot", "name": "BAGEL (7B MoT)", "level": "model", "family_id": null, "developer": [ "bytedance" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "other", "architecture_detail": "Mixture-of-Transformer-Experts (MoT): two transformer experts sharing self-attention at every layer", "params_total": 14000000000, "params_active": 7000000000, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-05", "release_date_basis": "inferred from page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BAGEL, a unified multimodal model from ByteDance Seed pretrained for both understanding and generation, here tracked for text-to-image generation.", "verified_on": "2026-10-11", "sources": [ "https://arxiv.org/abs/2505.14683" ], "unverified": [] }, { "id": "baguan", "name": "Baguan", "level": "model", "family_id": null, "developer": [ "alibaba-damo" ], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-05-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Weather model on WeatherBench 2 (2020).", "verified_on": "2026-10-10", "sources": [ "https://researchsites.withgoogle.com/weatherbench/deterministic/", "https://arxiv.org/abs/2505.13873" ], "unverified": [] }, { "id": "baichuan-ai-baichuan-m2-32b", "name": "Baichuan-M2-32B", "level": "model", "family_id": null, "developer": [ "baichuan-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 33000000000, "params_active": null, "params_checkpoint": null, "params_note": "33B parameters (BF16) per model card", "context_length": null, "release_date": "2025-09-02", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Chinese medical LLM fine-tuned from Qwen2.5-32B; HealthBench 60.1 (developer-reported).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/baichuan-inc/Baichuan-M2-32B" ], "unverified": [] }, { "id": "baichuan-baichuan-m3-235b", "name": "Baichuan-M3-235B", "level": "model", "family_id": null, "developer": [ "baichuan" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 235000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-02-09", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "multi-GPU node" ], "footprint_note": "GPTQ and FP8 quants exist", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Baichuan 235B model described as a medical LLM.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/baichuan-inc" ], "unverified": [ "summary", "licence" ] }, { "id": "baidu-ernie", "name": "ERNIE", "level": "family", "family_id": null, "developer": [ "baidu" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Ernie_Bot" ], "unverified": [] }, { "id": "baidu-ernie-4-0", "name": "ERNIE 4.0", "level": "model", "family_id": "baidu-ernie", "developer": [ "baidu" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Ernie_Bot" ], "unverified": [] }, { "id": "baidu-ernie-4-5", "name": "ERNIE 4.5", "level": "model", "family_id": "baidu-ernie", "developer": [ "baidu" ], "model_class": "multimodal LLM", "modalities_in": [], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "Heterogeneous multimodal MoE", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Ernie_Bot" ], "unverified": [] }, { "id": "baidu-ernie-4-5-0-3b", "name": "ERNIE-4.5-0.3B", "level": "model", "family_id": null, "developer": [ "baidu" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": 300000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small ERNIE 4.5 language model used as the decoder of PaddleOCR-VL-1.6.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2606.03264" ], "unverified": [] }, { "id": "baidu-ernie-5-0", "name": "ERNIE 5.0", "level": "model", "family_id": "baidu-ernie", "developer": [ "baidu" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Ernie_Bot" ], "unverified": [] }, { "id": "baidu-ernie-5-0-thinking-preview", "name": "ERNIE 5.0 Thinking Preview", "level": "variant", "family_id": "baidu-ernie", "developer": [ "baidu" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 128000, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/leaderboards/models" ], "unverified": [] }, { "id": "baidu-ernie-5-1", "name": "ERNIE 5.1", "level": "model", "family_id": "baidu-ernie", "developer": [ "baidu" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04-29", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Baidu's current ERNIE model; Wikipedia infobox lists an Apache licence (unverified for weights).", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Ernie_Bot", "https://arena.ai/leaderboard/chat/text" ], "unverified": [] }, { "id": "baidu-ernie-x1", "name": "ERNIE X1", "level": "model", "family_id": "baidu-ernie", "developer": [ "baidu" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Ernie_Bot" ], "unverified": [] }, { "id": "baidu-paddleocr-vl", "name": "PaddleOCR-VL", "level": "model", "family_id": null, "developer": [ "baidu" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "autoregressive", "architecture_detail": null, "params_total": 958600000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10", "release_date_basis": "announcement page", "weights_date": "2025-10-16", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First PaddleOCR-VL document VLM (v1).", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2510.14528", "https://huggingface.co/api/models/PaddlePaddle/PaddleOCR-VL" ], "unverified": [] }, { "id": "baidu-paddleocr-vl-1-5", "name": "PaddleOCR-VL-1.5", "level": "model", "family_id": null, "developer": [ "baidu" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "autoregressive", "architecture_detail": null, "params_total": 900000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01-29", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Baidu compact document VLM, parent of 1.6.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2606.03264", "https://huggingface.co/PaddlePaddle/PaddleOCR-VL-1.5" ], "unverified": [] }, { "id": "baidu-paddleocr-vl-1-6", "name": "PaddleOCR-VL-1.6", "level": "model", "family_id": null, "developer": [ "baidu" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "autoregressive", "architecture_detail": "Compact VLM: NaViT-style native-resolution encoder + adaptive MLP connector + ERNIE-4.5-0.3B decoder; PP-DocLayoutV3 layout stage", "params_total": 900000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-06-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU", "CPU server", "embedded GPU" ], "footprint_note": "Single GPU; small enough for CPU/edge with effort (judgement)", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Baidu compact document-parsing VLM; top of OmniDocBench v1.6 in its own paper.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2606.03264", "https://arxiv.org/pdf/2510.14528", "https://huggingface.co/PaddlePaddle/PaddleOCR-VL-1.6" ], "unverified": [] }, { "id": "baidu-rt-detr", "name": "RT-DETR", "level": "model", "family_id": null, "developer": [ "baidu", "peking-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image", "text" ], "architecture": "dense transformer", "architecture_detail": "Real-time DETR", "params_total": 43000000, "params_active": null, "params_checkpoint": null, "params_note": "43.0M for rtdetr_r50vd", "context_length": null, "release_date": "2023-04-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Baidu and Peking University real-time DETR ('DETRs Beat YOLOs on Real-time Object Detection'); ancestor of the D-FINE/DEIM line (unverified).", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2304.08069", "https://huggingface.co/api/models/PekingU/rtdetr_r50vd" ], "unverified": [] }, { "id": "bharatgen-param-1-2-9b", "name": "Param-1 2.9B", "level": "model", "family_id": null, "developer": [ "bharatgen" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 2900000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/BharatGenAI" ], "unverified": [] }, { "id": "bharatgen-param-1-7b", "name": "Param-1 7B", "level": "model", "family_id": null, "developer": [ "bharatgen" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 7000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/BharatGenAI" ], "unverified": [] }, { "id": "bharatgen-param2-17b-a2-4b-thinking", "name": "Param2-17B-A2.4B-Thinking", "level": "model", "family_id": null, "developer": [ "bharatgen" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 17000000000, "params_active": 2400000000, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-09", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BharatGen's 17B MoE thinking model; licence not verified.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/BharatGenAI" ], "unverified": [ "licence" ] }, { "id": "bigcode-starcoder2-15b", "name": "StarCoder2-15B", "level": "model", "family_id": null, "developer": [ "bigcode" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 15000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 16000, "release_date": "2024-02-29", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "bigcode-openrail-m-v1", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU", "laptop" ], "footprint_note": "T4/T5 estimate (unverified).", "runtimes": [], "languages_note": "Trained on The Stack v2, 600+ programming languages, 4T+ tokens", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BigCode legacy open code model; no 2026 successor found.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/bigcode/starcoder2-15b" ], "unverified": [ "deployment_tiers" ] }, { "id": "black-forest-labs-flux", "name": "FLUX", "level": "family", "family_id": null, "developer": [ "black-forest-labs" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Black Forest Labs' FLUX model line (FLUX.1, Kontext, FLUX.2, FLUX 3).", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Black_Forest_Labs", "https://datanorth.ai/news/black-forest-labs-releases-flux-3" ], "unverified": [] }, { "id": "black-forest-labs-flux-1", "name": "FLUX.1", "level": "family", "family_id": "black-forest-labs-flux", "developer": [ "black-forest-labs" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-08", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First FLUX generation (Aug 2024); Grok used FLUX in Aug 2024.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Black_Forest_Labs", "https://mlcommons.org/benchmarks/training/", "https://bfl.ai/announcements/24-08-01-bfl" ], "unverified": [] }, { "id": "black-forest-labs-flux-1-dev", "name": "FLUX.1 [dev]", "level": "model", "family_id": "black-forest-labs-flux-1", "developer": [ "black-forest-labs" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "FLUX.1 dev image model with a large LoRA ecosystem on Hugging Face (still trending, 15.5k likes).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/models?pipeline_tag=text-to-image&sort=trending", "https://en.wikipedia.org/wiki/Black_Forest_Labs" ], "unverified": [] }, { "id": "black-forest-labs-flux-1-kontext", "name": "FLUX.1 Kontext", "level": "model", "family_id": "black-forest-labs-flux", "developer": [ "black-forest-labs" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-05", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "FLUX image editing model (May 2025); Kontext Pro is used in Adobe Photoshop.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Black_Forest_Labs" ], "unverified": [] }, { "id": "black-forest-labs-flux-2", "name": "FLUX.2", "level": "family", "family_id": "black-forest-labs-flux", "developer": [ "black-forest-labs" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-11-25", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "FLUX.2 family (pro, flex, dev, klein) released 25 Nov 2025.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Black_Forest_Labs" ], "unverified": [] }, { "id": "black-forest-labs-flux-2-dev", "name": "FLUX.2 [dev]", "level": "model", "family_id": "black-forest-labs-flux-2", "developer": [ "black-forest-labs" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "flow matching", "architecture_detail": "Rectified flow transformer", "params_total": 32000000000, "params_active": null, "params_checkpoint": null, "params_note": "32B per S12 for the FLUX 2 Max comparison; dev size unverified", "context_length": null, "release_date": "2025-11-25", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights restricted", "licence": "flux-non-commercial", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open-weight non-commercial FLUX.2 image model; commercial use needs a licence from BFL.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Black_Forest_Labs", "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text", "https://huggingface.co/models?pipeline_tag=text-to-image&sort=trending", "https://datanorth.ai/news/qwen-releases-qwen-image-2-1" ], "unverified": [ "deployment_tiers" ] }, { "id": "black-forest-labs-flux-2-klein-4b", "name": "FLUX.2 [klein] 4B", "level": "model", "family_id": "black-forest-labs-flux-2", "developer": [ "black-forest-labs" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "flow matching", "architecture_detail": "Rectified flow transformer, 4-step distilled, Qwen3 8B text embedder", "params_total": 4000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-11-25", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Smallest FLUX.2 image model, Apache 2.0.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/black-forest-labs/FLUX.2-klein-9B", "https://en.wikipedia.org/wiki/Black_Forest_Labs", "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text" ], "unverified": [] }, { "id": "black-forest-labs-flux-2-klein-9b", "name": "FLUX.2 [klein] 9B", "level": "model", "family_id": "black-forest-labs-flux-2", "developer": [ "black-forest-labs" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "flow matching", "architecture_detail": "Rectified flow transformer, 4-step distilled, Qwen3 8B text embedder", "params_total": 9000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-11-25", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights restricted", "licence": "flux-non-commercial", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": "About 29 GB, RTX 4090 and above", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small distilled FLUX.2 image model under the FLUX Non-Commercial licence.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/black-forest-labs/FLUX.2-klein-9B", "https://en.wikipedia.org/wiki/Black_Forest_Labs", "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text" ], "unverified": [] }, { "id": "black-forest-labs-flux-2-max", "name": "FLUX 2 Max", "level": "model", "family_id": "black-forest-labs-flux-2", "developer": [ "black-forest-labs" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "FLUX.2 model compared in Qwen's vendor benchmark chart.", "verified_on": "2026-10-10", "sources": [ "https://datanorth.ai/news/qwen-releases-qwen-image-2-1" ], "unverified": [] }, { "id": "black-forest-labs-flux-3-image", "name": "FLUX 3 Image", "level": "model", "family_id": "black-forest-labs-flux", "developer": [ "black-forest-labs" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": "Unified multimodal model building on BFL Self-Flow", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-07-23", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "gated preview", "API" ], "deployment_tiers": [], "footprint_note": "Cloud only at launch; open-weight faster versions promised later in 2026", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Black Forest Labs' unified image model announced 23 Jul 2026 (AA lists Oct 2026), proprietary at launch.", "verified_on": "2026-10-10", "sources": [ "https://datanorth.ai/news/black-forest-labs-releases-flux-3", "https://en.wikipedia.org/wiki/Black_Forest_Labs", "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text" ], "unverified": [] }, { "id": "black-forest-labs-flux-3-video", "name": "FLUX 3 Video", "level": "model", "family_id": "black-forest-labs-flux", "developer": [ "black-forest-labs" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": "Unified (Self-Flow)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-07-23", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "gated preview", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BFL's first video model, early access from 23 Jul 2026, up to 20 s with synchronised audio. Vendor claims 77% and 93% preference against Runway Gen-4.5 and Luma Ray 3.2.", "verified_on": "2026-10-10", "sources": [ "https://datanorth.ai/news/black-forest-labs-releases-flux-3", "https://artificialanalysis.ai/video/leaderboard/text-to-video", "https://arena.ai/leaderboard/text-to-video" ], "unverified": [] }, { "id": "black-forest-labs-flux1-1-pro", "name": "FLUX1.1 [pro]", "level": "model", "family_id": "black-forest-labs-flux-1", "developer": [ "black-forest-labs" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-10", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "BFL API endpoint flux-pro-1.1, USD 0.04 per image", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "FLUX1.1 [pro] is a Black Forest Labs text-to-image model served through the BFL API.", "verified_on": "2026-10-11", "sources": [ "https://docs.bfl.ai/flux_models/flux_1_1_pro", "https://artificialanalysis.ai/image/leaderboard/text-to-image" ], "unverified": [] }, { "id": "blip3-o-8b", "name": "BLIP3-o (8B)", "level": "model", "family_id": null, "developer": [ "salesforce" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "hybrid", "architecture_detail": "Frozen Qwen2.5-VL-7B-Instruct backbone with a diffusion transformer for image generation", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Frozen Qwen2.5-VL-7B-Instruct backbone plus about 1.4B trainable diffusion transformer parameters", "context_length": null, "release_date": "2025-05-14", "release_date_basis": "inferred from page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BLIP3-o 8B, a unified multimodal model for image understanding and generation from Salesforce Research and academic partners.", "verified_on": "2026-10-11", "sources": [ "https://arxiv.org/abs/2505.09568" ], "unverified": [] }, { "id": "blip3o-next-grpo-geneval-3b", "name": "BLIP3o-NEXT-GRPO-GenEval (3B)", "level": "model", "family_id": null, "developer": [ "salesforce" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "hybrid", "architecture_detail": "Qwen3 autoregressive model paired with a SANA1.5 diffusion transformer; GRPO-trained variant targeted at GenEval", "params_total": 3000000000, "params_active": null, "params_checkpoint": null, "params_note": "About 3B per the paper", "context_length": null, "release_date": "2025-10-17", "release_date_basis": "inferred from page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BLIP3o-NEXT 3B variant trained with GRPO for GenEval: a native image generation model combining an autoregressive model with a diffusion transformer.", "verified_on": "2026-10-11", "sources": [ "https://arxiv.org/abs/2510.15857" ], "unverified": [] }, { "id": "bloomberg-bloomberggpt", "name": "BloombergGPT", "level": "model", "family_id": null, "developer": [ "bloomberg" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": "Trained from scratch", "params_total": 50000000000, "params_active": null, "params_checkpoint": null, "params_note": "363B finance tokens plus 345B general tokens.", "context_length": null, "release_date": "2023-03", "release_date_basis": "notes", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "internal only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "BloombergGPT is a 50 billion parameter language model trained on a wide range of financial data, for financial NLP tasks.", "verified_on": "2026-10-10", "sources": [ "https://www.bloomberg.com/company/press/bloomberggpt-50-billion-parameter-llm", "https://arxiv.org/abs/2303.17564", "https://arxiv.org/html/2305.05862v2" ], "unverified": [] }, { "id": "bo-wang-lab-medsam", "name": "MedSAM", "level": "model", "family_id": null, "developer": [ "bo-wang-lab" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": null, "architecture_detail": "Initialized from the SAM ViT-B checkpoint", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Medical image segmentation model adapted from SAM, published in Nature Communications 15:654 (2024). Developer location (Toronto) unverified.", "verified_on": "2026-10-10", "sources": [ "https://github.com/bowang-lab/MedSAM" ], "unverified": [] }, { "id": "bo-wang-lab-medsam2", "name": "MedSAM2", "level": "model", "family_id": null, "developer": [ "bo-wang-lab" ], "model_class": "vision", "modalities_in": [ "image", "3D", "video" ], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-04-07", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Successor to MedSAM for 3D and video segmentation, released 7 Apr 2025.", "verified_on": "2026-10-10", "sources": [ "https://github.com/bowang-lab/MedSAM" ], "unverified": [] }, { "id": "boltz-boltz-1", "name": "Boltz-1", "level": "model", "family_id": null, "developer": [ "boltz", "mit" ], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-11", "release_date_basis": "secondary report", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Predecessor of Boltz-2.", "verified_on": "2026-10-10", "sources": [ "https://scalingbiotech.substack.com/p/open-source-protein-models-openfold", "https://intuitionlabs.ai/articles/open-source-ai-drug-discovery-boltz-pfizer", "https://github.com/jwohlwend/boltz", "https://par.nsf.gov/servlets/purl/10585253" ], "unverified": [] }, { "id": "boltz-boltz-2", "name": "Boltz-2", "level": "model", "family_id": null, "developer": [ "boltz", "mit", "recursion" ], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "diffusion", "architecture_detail": "AF3-style co-folding + binding affinity head", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-06", "release_date_basis": "secondary report", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Co-folding and binding-affinity model; vendor claims it approaches FEP accuracy at about 1000x speed; evaluated on FEP+, CASP16, MF-PCBA.", "verified_on": "2026-10-10", "sources": [ "https://scalingbiotech.substack.com/p/open-source-protein-models-openfold", "https://github.com/jwohlwend/boltz", "https://intuitionlabs.ai/articles/open-source-ai-drug-discovery-boltz-pfizer", "https://boltz.com/boltz2", "https://par.nsf.gov/servlets/purl/10585253" ], "unverified": [] }, { "id": "boltz-boltzgen", "name": "BoltzGen", "level": "model", "family_id": null, "developer": [ "boltz" ], "model_class": "scientific", "modalities_in": [], "modalities_out": [ "molecule" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Boltz design model; its licence is described only as permissive.", "verified_on": "2026-10-10", "sources": [ "https://intuitionlabs.ai/articles/open-source-ai-drug-discovery-boltz-pfizer", "https://github.com/jwohlwend/boltz" ], "unverified": [] }, { "id": "boson-higgs-tts-3-4b", "name": "Boson higgs-tts-3-4b", "level": "model", "family_id": null, "developer": [], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open TTS model trending on Hugging Face (2025 to 2026).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/models?pipeline_tag=text-to-speech&sort=trending" ], "unverified": [] }, { "id": "breezeblue-breeze-tts-2", "name": "Breeze TTS 2", "level": "model", "family_id": null, "developer": [ "breezeblue" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": "Text encoder plus backbone plus depth decoder plus codec; Qwen3-TTS tokenizer", "params_total": 3000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-08-25", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "breezeblue-research-non-commercial", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU", "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": "English, Chinese only", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BreezeBlue open TTS model, the only open model in the Artificial Analysis TTS top 25.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/BreezeBlue/Breeze-TTS-2", "https://artificialanalysis.ai/text-to-speech/leaderboard", "https://huggingface.co/BreezeBlue" ], "unverified": [] }, { "id": "bytedance-bagel", "name": "BAGEL (BAGEL-7B-MoT)", "level": "model", "family_id": null, "developer": [ "bytedance" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text", "image" ], "architecture": "MoE", "architecture_detail": "Mixture-of-Transformer-Experts (MoT) with pixel-level and semantic-level image encoders; next group of token prediction", "params_total": 14000000000, "params_active": 7000000000, "params_checkpoint": null, "params_note": "Card: 7B active, 14B total; safetensors metadata lists 15B", "context_length": null, "release_date": "2025-05-20", "release_date_basis": "inferred from page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ByteDance Seed unified multimodal model for image understanding, text-to-image generation and image editing, released with open weights.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/ByteDance-Seed/BAGEL-7B-MoT", "https://artificialanalysis.ai/image/leaderboard/text-to-image" ], "unverified": [] }, { "id": "bytedance-dola-seed-2-0-pro", "name": "Dola-Seed 2.0 Pro", "level": "model", "family_id": null, "developer": [ "bytedance" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://seed.bytedance.com/en/", "https://arena.ai/leaderboard/chat/text" ], "unverified": [] }, { "id": "bytedance-doubao-seed-code", "name": "Doubao Seed Code", "level": "model", "family_id": null, "developer": [ "bytedance" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 256000, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://seed.bytedance.com/en/", "https://artificialanalysis.ai/leaderboards/models" ], "unverified": [] }, { "id": "bytedance-latentsync-1-6", "name": "LatentSync 1.6", "level": "model", "family_id": null, "developer": [ "bytedance" ], "model_class": "avatar", "modalities_in": [ "audio", "video" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": "Audio-conditioned latent diffusion U-Net, 512 px", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "openrail-plus-plus", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ByteDance open lip-sync model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/ByteDance/LatentSync-1.6", "https://en.wikipedia.org/wiki/ByteDance" ], "unverified": [] }, { "id": "bytedance-protenix", "name": "Protenix", "level": "family", "family_id": null, "developer": [ "bytedance" ], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "other", "architecture_detail": "AF3 reproduction", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-11-08", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ByteDance AF3 reproduction family.", "verified_on": "2026-10-10", "sources": [ "https://neurosnap.ai/blog/post/from-alphafold3-to-protenix-making-biomolecular-modeling-more-practical/6772f2644fd5e47a17836a41", "https://github.com/bytedance/Protenix" ], "unverified": [] }, { "id": "bytedance-protenix-mini", "name": "Protenix-Mini / Mini+", "level": "model", "family_id": "bytedance-protenix", "developer": [ "bytedance" ], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "other", "architecture_detail": "Compact architecture, few-step diffusion, switchable pLM", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-07-16", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Compact Protenix (Mini+ arXiv 13 Oct 2025); code Apache-2.0.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2507.11839", "https://arxiv.org/abs/2510.12842" ], "unverified": [] }, { "id": "bytedance-protenix-v1", "name": "Protenix-v1", "level": "model", "family_id": "bytedance-protenix", "developer": [ "bytedance" ], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "other", "architecture_detail": "AF3 reproduction with MSA, RNA MSA and templates", "params_total": 368000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-02-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open AF3 reproduction (protenix_base_default_v1.0.0); vendor claims it outperforms AlphaFold3 at the same training cutoff. Also protenix_base_20250630_v1.0.0 with data cutoff 30 Jun 2025.", "verified_on": "2026-10-10", "sources": [ "https://github.com/bytedance/Protenix", "https://pypi.org/pypi/protenix/json", "https://github.com/bytedance/Protenix" ], "unverified": [] }, { "id": "bytedance-protenix-v2", "name": "Protenix-v2", "level": "model", "family_id": "bytedance-protenix", "developer": [ "bytedance" ], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "other", "architecture_detail": null, "params_total": 464000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "internal only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Protenix v2 with gains on antibody-antigen prediction; weights proprietary and confidential.", "verified_on": "2026-10-10", "sources": [ "https://github.com/bytedance/Protenix" ], "unverified": [] }, { "id": "bytedance-seed-1-6-thinking", "name": "Seed1.6-Thinking", "level": "model", "family_id": "bytedance-seed1-6", "developer": [ "bytedance" ], "model_class": "reasoning", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "Sparse MoE building on Seed1.5; Seed1.6 Base has 23B active and 230B total parameters", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Blog gives 230B total, 23B active for Seed1.6 Base only, not for post-trained variants", "context_length": 256000, "release_date": "2025-06", "release_date_basis": "inferred from page", "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ByteDance Seed variant of Seed1.6 trained for extended reasoning over text and images, served through the Volcano Engine API.", "verified_on": "2026-10-11", "sources": [ "https://seed.bytedance.com/en/blog/introduction-to-techniques-used-in-seed1-6" ], "unverified": [ "release_date" ] }, { "id": "bytedance-seed-coder-8b-base", "name": "Seed-Coder-8B-Base", "level": "model", "family_id": null, "developer": [ "bytedance" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/ByteDance-Seed/Seed-Coder-8B-Instruct" ], "unverified": [] }, { "id": "bytedance-seed-coder-8b-instruct", "name": "Seed-Coder-8B-Instruct", "level": "model", "family_id": null, "developer": [ "bytedance" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 32000, "release_date": "2025-06-04", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "laptop" ], "footprint_note": "T5 estimate (unverified). Completion-class only.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ByteDance Seed 8B open code model built on the Seed-Coder base.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/ByteDance-Seed/Seed-Coder-8B-Instruct" ], "unverified": [ "deployment_tiers" ] }, { "id": "bytedance-seed1-6", "name": "Seed1.6", "level": "family", "family_id": null, "developer": [ "bytedance" ], "model_class": "multimodal LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-06-25", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Seed1.6 is the ByteDance Seed general-purpose model series with adaptive deep thinking and multimodal understanding.", "verified_on": "2026-10-11", "sources": [ "https://seed.bytedance.com/en/blog/introduction-to-techniques-used-in-seed1-6" ], "unverified": [] }, { "id": "bytedance-seed2-1", "name": "Seed2.1", "level": "model", "family_id": null, "developer": [ "bytedance" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ByteDance Seed agent model, undated.", "verified_on": "2026-10-10", "sources": [ "https://seed.bytedance.com/en/" ], "unverified": [] }, { "id": "bytedance-seedance", "name": "Seedance", "level": "family", "family_id": null, "developer": [ "bytedance" ], "model_class": "video generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ByteDance Seed video model line.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Seedance" ], "unverified": [] }, { "id": "bytedance-seedance-1-0", "name": "Seedance 1.0", "level": "model", "family_id": "bytedance-seedance", "developer": [ "bytedance" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-06", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First Seedance model, Jun 2025 (Pro variant on VBench-2.0).", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Seedance", "https://huggingface.co/spaces/Vchitect/VBench_Leaderboard" ], "unverified": [] }, { "id": "bytedance-seedance-1-5-pro", "name": "Seedance 1.5 Pro", "level": "model", "family_id": "bytedance-seedance", "developer": [ "bytedance" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier Seedance model.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Seedance" ], "unverified": [] }, { "id": "bytedance-seedance-2-0", "name": "Seedance 2.0", "level": "model", "family_id": "bytedance-seedance", "developer": [ "bytedance" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-02", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ByteDance video model; drew a Disney cease-and-desist (13 Feb 2026) and US senators urged shutdown (16 Mar 2026).", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Seedance", "https://artificialanalysis.ai/video/leaderboard/text-to-video", "https://arena.ai/leaderboard/text-to-video" ], "unverified": [] }, { "id": "bytedance-seedance-2-5", "name": "Seedance 2.5", "level": "model", "family_id": "bytedance-seedance", "developer": [ "bytedance" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-07-31", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ByteDance media or voice model, noted in passing in the research notes.", "verified_on": "2026-10-10", "sources": [ "https://seed.bytedance.com/en/", "https://en.wikipedia.org/wiki/Seedance", "https://artificialanalysis.ai/video/leaderboard/text-to-video", "https://arena.ai/leaderboard/text-to-video" ], "unverified": [] }, { "id": "bytedance-seedrealtime", "name": "SeedRealtime", "level": "model", "family_id": null, "developer": [ "bytedance" ], "model_class": "speech synthesis", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ByteDance media or voice model, noted in passing in the research notes.", "verified_on": "2026-10-10", "sources": [ "https://seed.bytedance.com/en/" ], "unverified": [] }, { "id": "bytedance-seedream", "name": "Seedream", "level": "family", "family_id": null, "developer": [ "bytedance" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ByteDance Seed image model line.", "verified_on": "2026-10-10", "sources": [ "https://digg.com/ai/19tzlk11", "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text" ], "unverified": [] }, { "id": "bytedance-seedream-4-0", "name": "Seedream 4.0", "level": "model", "family_id": "bytedance-seedream", "developer": [ "bytedance" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-09", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ByteDance image model, Sep 2025.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text", "https://digg.com/ai/19tzlk11" ], "unverified": [] }, { "id": "bytedance-seedream-4-5", "name": "Seedream 4.5", "level": "model", "family_id": "bytedance-seedream", "developer": [ "bytedance" ], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-12", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ByteDance Seed image generation and editing model with better multi-image editing, reference consistency and dense text rendering than Seedream 4.0.", "verified_on": "2026-10-11", "sources": [ "https://seed.bytedance.com/en/seedream4_5", "https://artificialanalysis.ai/image/leaderboard/text-to-image", "https://artificialanalysis.ai/image/model-families/Seedream" ], "unverified": [] }, { "id": "bytedance-seedream-5-0-lite", "name": "Seedream 5.0 Lite", "level": "model", "family_id": "bytedance-seedream", "developer": [ "bytedance" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Seedream model described as first with online search (search title only).", "verified_on": "2026-10-10", "sources": [ "https://digg.com/ai/19tzlk11" ], "unverified": [ "summary" ] }, { "id": "bytedance-seedream-5-0-pro", "name": "Seedream 5.0 Pro", "level": "model", "family_id": "bytedance-seedream", "developer": [ "bytedance" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-07", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app", "API" ], "deployment_tiers": [], "footprint_note": "Dreamina / Jimeng apps and API; OpenRouter slug seedream-5-0-pro-20260812 (unverified)", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ByteDance media or voice model, noted in passing in the research notes.", "verified_on": "2026-10-10", "sources": [ "https://seed.bytedance.com/en/", "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text", "https://arena.ai/leaderboard/text-to-image", "https://digg.com/ai/19tzlk11", "https://en.wikipedia.org/wiki/ByteDance" ], "unverified": [] }, { "id": "cad-ose", "name": "CAD OSE", "level": "model", "family_id": null, "developer": [], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Detector on the NAB scoreboard.", "verified_on": "2026-10-10", "sources": [ "https://github.com/numenta/NAB" ], "unverified": [] }, { "id": "caev2", "name": "CAEv2", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": null, "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Backbone in LW-DETR that RF-DETR replaces with DINOv2.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2511.09554v2" ], "unverified": [] }, { "id": "carnegie-mellon-university-moment-1-large", "name": "MOMENT (MOMENT-1-large)", "level": "model", "family_id": null, "developer": [ "carnegie-mellon-university" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "dense transformer", "architecture_detail": "T5-encoder masked; pre-trained on the Time series Pile", "params_total": 346400000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-02-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "General-purpose open time-series model for forecasting, classification and anomaly detection; not submitted to GIFT-Eval. Last PyPI momentfm 0.1.4 on 19 Mar 2025.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2402.03885v3", "https://huggingface.co/spaces/Salesforce/GIFT-Eval", "https://huggingface.co/api/models/AutonLab/MOMENT-1-large", "https://pypi.org/pypi/momentfm/json", "https://huggingface.co/AutonLab/MOMENT-1-large", "https://arxiv.org/pdf/2602.01359" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "cartesia-sonic", "name": "Sonic", "level": "family", "family_id": null, "developer": [ "cartesia" ], "model_class": "speech synthesis", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Cartesia Sonic TTS models.", "verified_on": "2026-10-10", "sources": [ "https://cartesia.ai/sonic" ], "unverified": [] }, { "id": "cartesia-sonic-3-5", "name": "Sonic 3.5", "level": "model", "family_id": "cartesia-sonic", "developer": [ "cartesia" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-05", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier Sonic model.", "verified_on": "2026-10-10", "sources": [ "https://cartesia.ai/sonic" ], "unverified": [] }, { "id": "cartesia-sonic-3-6", "name": "Sonic 3.6", "level": "model", "family_id": "cartesia-sonic", "developer": [ "cartesia" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "SSM", "architecture_detail": "State Space Model", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-08", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "Cloud or on-prem deployment offered", "runtimes": [], "languages_note": "44 languages", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Cartesia GA TTS model with sub-90 ms latency, 4th on the Artificial Analysis TTS arena.", "verified_on": "2026-10-10", "sources": [ "https://cartesia.ai/sonic", "https://artificialanalysis.ai/text-to-speech/leaderboard", "https://docs.cartesia.ai/build-with-cartesia/tts-models/latest" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "chai-discovery-chai-1", "name": "Chai-1", "level": "model", "family_id": null, "developer": [ "chai-discovery" ], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "other", "architecture_detail": "Multi-modal structure prediction: proteins, small molecules, DNA, RNA, glycosylations", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-11-27", "release_date_basis": "registry date", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Co-folding model, Apache-2.0 for code and weights (academic and commercial); chai_lab 0.6.1 on 18 Mar 2025.", "verified_on": "2026-10-10", "sources": [ "https://github.com/chaidiscovery/chai-lab", "https://huggingface.co/api/models/chaidiscovery/chai-1", "https://pypi.org/pypi/chai_lab/json" ], "unverified": [] }, { "id": "chai-discovery-chai-2", "name": "Chai-2", "level": "model", "family_id": null, "developer": [ "chai-discovery" ], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "other", "architecture_detail": "Multimodal generative model for de novo antibody and miniprotein design", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-07-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Chai Discovery antibody design model; 16% hit rate in fully de novo antibody design (vendor), at least one hit for 50% of 52 targets; 68% wet-lab success in miniprotein design.", "verified_on": "2026-10-10", "sources": [ "https://www.biorxiv.org/content/10.1101/2025.07.05.663018v1" ], "unverified": [] }, { "id": "clay-foundation-clay", "name": "Clay", "level": "model", "family_id": null, "developer": [ "clay-foundation" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": "MAE expanded visual transformer; tags geovit + DOFA", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-12-29", "release_date_basis": "registry date", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open EO foundation model (v1.5 era); docs CC-BY-4.0, embeddings ODC-BY; latest GitHub tag v0.0.1. Used as baseline in the AlphaEarth paper.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/made-with-clay/Clay", "https://arxiviq.substack.com/p/alphaearth-foundations-an-embedding" ], "unverified": [ "release_date" ] }, { "id": "cnpc-kunlun-large-model", "name": "Kunlun Large Model", "level": "model", "family_id": null, "developer": [ "cnpc" ], "model_class": "vision", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": "Private cloud (unverified)", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "CNPC model on Huawei Pangu for pipeline porosity and micro-crack detection; vendor claims about 40% higher defect identification efficiency and 25% less manual work.", "verified_on": "2026-10-10", "sources": [ "https://www.intelligentcio.com/apac/2025/06/24/cross-industry-use-cases-for-huaweis-pangu-models-5-5-showcased/" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "cognition-swe", "name": "SWE (Devin)", "level": "family", "family_id": null, "developer": [ "cognition" ], "model_class": "code", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://www.orcarouter.ai/fr/blog/swe-2" ], "unverified": [] }, { "id": "cognition-swe-1-6-preview", "name": "SWE-1.6 preview", "level": "model", "family_id": "cognition-swe", "developer": [ "cognition" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://www.orcarouter.ai/fr/blog/swe-2" ], "unverified": [] }, { "id": "cognition-swe-1-7", "name": "SWE-1.7", "level": "model", "family_id": "cognition-swe", "developer": [ "cognition" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://www.orcarouter.ai/fr/blog/swe-2" ], "unverified": [] }, { "id": "cognition-swe-2", "name": "SWE-2", "level": "model", "family_id": "cognition-swe", "developer": [ "cognition" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Not published; base model 2.8T (Kimi K3), base context 1M", "context_length": null, "release_date": "2026-09-10", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app" ], "deployment_tiers": [], "footprint_note": "Devin Desktop/CLI only; no API, no weights.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Cognition's Devin coding model, post-trained from Kimi K3.", "verified_on": "2026-10-10", "sources": [ "https://www.orcarouter.ai/fr/blog/swe-2" ], "unverified": [] }, { "id": "cogvideox-i2v", "name": "CogVideoX-I2V", "level": "model", "family_id": null, "developer": [ "zhipu-zai" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-09-16", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Model scored in the WorldScore paper table.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2504.00983v2", "https://huggingface.co/THUDM/CogVideoX-5b-I2V" ], "unverified": [] }, { "id": "cohere-aya-expanse-32b", "name": "Aya Expanse 32B", "level": "model", "family_id": null, "developer": [ "cohere" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 32000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 128000, "release_date": "2024-10-23", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "23 languages including Indonesian (vendor)", "indonesian_support": "listed by developer", "regional_languages": [ "Indonesian" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Cohere's multilingual 32B model covering 23 languages including Indonesian.", "verified_on": "2026-10-10", "sources": [ "https://docs.cohere.com/docs/models", "https://huggingface.co/CohereLabs/aya-expanse-32b" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "cohere-command-a-03-2025", "name": "command-a-03-2025", "level": "model", "family_id": null, "developer": [ "cohere" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://arena.ai/leaderboard/chat/text" ], "unverified": [] }, { "id": "cohere-command-a-plus", "name": "Command A+ (command-a-plus-05-2026)", "level": "model", "family_id": null, "developer": [ "cohere" ], "model_class": "multimodal LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Context conflict: 128k (64k output) per Cohere docs vs 192k per AA", "context_length": null, "release_date": "2026-05", "release_date_basis": "inferred from page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Not listed", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Cohere's Command A+ model, text and image input.", "verified_on": "2026-10-10", "sources": [ "https://docs.cohere.com/docs/models", "https://artificialanalysis.ai/leaderboards/models" ], "unverified": [] }, { "id": "cohere-command-a-reasoning", "name": "Command A Reasoning (08-2025)", "level": "model", "family_id": null, "developer": [ "cohere" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 256000, "release_date": "2025-08", "release_date_basis": "inferred from page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": "Max output 32k.", "runtimes": [], "languages_note": "23 languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://docs.cohere.com/docs/models" ], "unverified": [] }, { "id": "cohere-command-a-translate", "name": "Command A Translate", "level": "model", "family_id": null, "developer": [ "cohere" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "23 languages including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Indonesian" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Cohere translation model listing Indonesian among 23 languages.", "verified_on": "2026-10-10", "sources": [ "https://docs.cohere.com/docs/models" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "cohere-command-r-mar-24", "name": "Command R (Mar '24)", "level": "model", "family_id": null, "developer": [ "cohere", "cohere-labs" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "Auto-regressive optimised transformer with SFT and preference training", "params_total": 35000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 128000, "release_date": "2024-03-11", "release_date_basis": "announcement page", "weights_date": "2024-03-11", "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Optimised for English, French, Spanish, Italian, German, Brazilian Portuguese, Japanese, Korean, Simplified Chinese and Arabic; pre-training also included Russian, Polish, Turkish, Vietnamese, Dutch, Czech, Indonesian, Ukrainian, Romanian, Greek, Hindi, Hebrew and Persian", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Command R is a 35B Cohere research release optimised for reasoning, summarisation, multilingual generation and retrieval-augmented generation with citations and tool use.", "verified_on": "2026-10-11", "sources": [ "https://cohere.com/blog/command-r", "https://huggingface.co/CohereLabs/c4ai-command-r-v01", "https://huggingface.co/api/models/CohereForAI/c4ai-command-r-v01" ], "unverified": [ "indonesian_measured" ] }, { "id": "cohere-command-r-plus-apr-24", "name": "Command R+ (Apr '24)", "level": "model", "family_id": null, "developer": [ "cohere", "cohere-labs" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "Auto-regressive optimised transformer with SFT and preference training", "params_total": 104000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 128000, "release_date": "2024-04-04", "release_date_basis": "announcement page", "weights_date": "2024-04-03", "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "API", "cloud marketplace", "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Optimised for English, French, Spanish, Italian, German, Brazilian Portuguese, Japanese, Korean, Simplified Chinese and Arabic; pre-training also included Russian, Polish, Turkish, Vietnamese, Dutch, Czech, Indonesian, Ukrainian, Romanian, Greek, Hindi, Hebrew and Persian", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Command R+ is a 104B Cohere model optimised for retrieval-augmented generation and multi-step tool use, released as open weights for research under CC-BY-NC and first on Microsoft Azure.", "verified_on": "2026-10-11", "sources": [ "https://cohere.com/blog/command-r-plus-microsoft-azure", "https://huggingface.co/CohereLabs/c4ai-command-r-plus", "https://huggingface.co/api/models/CohereForAI/c4ai-command-r-plus" ], "unverified": [ "indonesian_measured" ] }, { "id": "cohere-labs-tiny-aya", "name": "Tiny Aya", "level": "family", "family_id": null, "developer": [ "cohere-labs" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-03-12", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Cohere Labs' 3.35B multilingual small models, non-commercial (CC-BY-NC-4.0 plus AUP); Indonesian is named on the card.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/CohereLabs", "https://huggingface.co/CohereLabs/tiny-aya-global" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "cohere-labs-tiny-aya-base", "name": "Tiny Aya base", "level": "variant", "family_id": "cohere-labs-tiny-aya", "developer": [ "cohere-labs" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 3350000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 8000, "release_date": "2026-03-12", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "laptop", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/CohereLabs" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "cohere-labs-tiny-aya-earth", "name": "Tiny Aya earth", "level": "variant", "family_id": "cohere-labs-tiny-aya", "developer": [ "cohere-labs" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 3350000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 8000, "release_date": "2026-03-12", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "laptop", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/CohereLabs" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "cohere-labs-tiny-aya-fire", "name": "Tiny Aya fire", "level": "variant", "family_id": "cohere-labs-tiny-aya", "developer": [ "cohere-labs" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 3350000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 8000, "release_date": "2026-03-12", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "laptop", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/CohereLabs" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "cohere-labs-tiny-aya-global", "name": "Tiny Aya global", "level": "variant", "family_id": "cohere-labs-tiny-aya", "developer": [ "cohere-labs" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 3350000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 8000, "release_date": "2026-03-12", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "laptop", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/CohereLabs", "https://huggingface.co/CohereLabs/tiny-aya-global" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "cohere-labs-tiny-aya-water", "name": "Tiny Aya water", "level": "variant", "family_id": "cohere-labs-tiny-aya", "developer": [ "cohere-labs" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 3350000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 8000, "release_date": "2026-03-12", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "laptop", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/CohereLabs", "https://leaderboard.sea-lion.ai/detailed/ID" ], "unverified": [ "indonesia_hosting" ], "indonesian_measured": [ { "benchmark": "SEA-HELM Indonesian leaderboard (ID aggregate, first score column of the Indonesian Competency and Task Scores table)", "score": "33.23 (95% CI -1.13 to +1.13)", "metric": "score 0 to 100, average of 2000 bootstraps with 95% CI", "run_by": "AI Singapore (SEA-HELM leaderboard maintainer)", "url": "https://leaderboard.sea-lion.ai/detailed/ID" } ] }, { "id": "cohere-north", "name": "North models", "level": "family", "family_id": null, "developer": [ "cohere" ], "model_class": "code", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Cohere next-generation North model line, inaugurated by North Mini Code.", "verified_on": "2026-10-11", "sources": [ "https://cohere.com/blog/north-mini-code" ], "unverified": [ "model_class" ] }, { "id": "cohere-north-mini-code", "name": "North Mini Code", "level": "model", "family_id": "cohere-north", "developer": [ "cohere" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "Decoder-only sparse MoE; interleaved sliding-window attention with RoPE and global attention without positional embeddings", "params_total": 30000000000, "params_active": 3000000000, "params_checkpoint": null, "params_note": null, "context_length": 256000, "release_date": "2026-06-09", "release_date_basis": "announcement page", "weights_date": "2026-06-09", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "API", "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": "Runs on 1 H100 at FP8 or FP4; 64K max output; weights in bf16, fp8 and w4a16", "runtimes": [ "vllm" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Cohere first agentic coding model, an open-weights 30B MoE for code generation, agentic software engineering and terminal tasks.", "verified_on": "2026-10-11", "sources": [ "https://cohere.com/blog/north-mini-code", "https://huggingface.co/CohereLabs/North-Mini-Code-1.0", "https://openrouter.ai/cohere/north-mini-code" ], "unverified": [] }, { "id": "cohere-transcribe", "name": "Cohere Transcribe (cohere-transcribe-03-2026)", "level": "model", "family_id": null, "developer": [ "cohere" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "autoregressive", "architecture_detail": "Conformer + LLM decoder (pattern, unverified for this model)", "params_total": 2000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-03-26", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Cohere open ASR model; led the Open ASR Leaderboard in March 2026.", "verified_on": "2026-10-10", "sources": [ "https://select.keenable.ai/r/speech-recognition-accuracy-leaders-2022-2026/JbURQjlVooLQ6J10NkbRVw", "https://benchmarklist.com/benchmarks/open_asr_leaderboard/" ], "unverified": [ "architecture_detail" ] }, { "id": "conv-ae", "name": "Conv-AE", "level": "model", "family_id": null, "developer": [], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Convolutional autoencoder anomaly detector, top of the SKAB outlier leaderboard.", "verified_on": "2026-10-10", "sources": [ "https://github.com/waico/SKAB" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "coqui-xtts-v2", "name": "XTTS-v2", "level": "model", "family_id": null, "developer": [ "coqui" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "coqui-public-model-license", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": "17 languages, not Indonesian", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Coqui open TTS model with 6.7M downloads.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/coqui/XTTS-v2" ], "unverified": [] }, { "id": "ctranspath", "name": "CTransPath", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Pathology model used as a comparison in the Virchow study (AUC 0.907); developer not stated in these sections.", "verified_on": "2026-10-10", "sources": [ "https://www.nature.com/articles/s41591-024-03141-0" ], "unverified": [] }, { "id": "databricks-dbrx", "name": "DBRX", "level": "model", "family_id": null, "developer": [ "databricks" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "Decoder-only fine-grained MoE: 16 experts, 4 active; RoPE, GLU, GQA; trained on 12T tokens of text and code", "params_total": 132000000000, "params_active": 36000000000, "params_checkpoint": null, "params_note": null, "context_length": 32000, "release_date": "2024-03-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "databricks-open-model-license", "access_routes": [ "API", "cloud marketplace", "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "DBRX is an open general-purpose mixture-of-experts LLM from Databricks (Base and Instruct), 36B active of 132B parameters.", "verified_on": "2026-10-11", "sources": [ "https://www.databricks.com/blog/introducing-dbrx-new-state-art-open-llm", "https://www.databricks.com/legal/open-model-license" ], "unverified": [] }, { "id": "datadog-toto-1-0", "name": "Toto 1.0", "level": "model", "family_id": null, "developer": [ "datadog" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Predecessor of Toto 2.0.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2605.20119" ], "unverified": [] }, { "id": "datadog-toto-2-0", "name": "Toto 2.0", "level": "family", "family_id": null, "developer": [ "datadog" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "dense transformer", "architecture_detail": "Decoder transformer (unverified detail)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "4M, 22M, 313M, 1B, 2.5B", "context_length": null, "release_date": "2026-06-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server", "embedded GPU", "single datacentre GPU" ], "footprint_note": "4M/22M CPU/edge; 2.5B single GPU (judgement)", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Datadog time-series foundation model family, Apache-2.0 for all five base checkpoints.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2605.20119", "https://huggingface.co/spaces/Salesforce/GIFT-Eval" ], "unverified": [] }, { "id": "datadog-toto-2-0-1b", "name": "Toto 2.0 1B", "level": "variant", "family_id": "datadog-toto-2-0", "developer": [ "datadog" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 1000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-06-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Toto 2.0 size variant.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2605.20119" ], "unverified": [] }, { "id": "datadog-toto-2-0-2-5b", "name": "Toto 2.0 2.5B", "level": "variant", "family_id": "datadog-toto-2-0", "developer": [ "datadog" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 2500000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-06-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Toto 2.0 size variant.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2605.20119" ], "unverified": [] }, { "id": "datadog-toto-2-0-22m", "name": "Toto 2.0 22M", "level": "variant", "family_id": "datadog-toto-2-0", "developer": [ "datadog" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 22000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-06-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Toto 2.0 size variant.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2605.20119" ], "unverified": [] }, { "id": "datadog-toto-2-0-313m", "name": "Toto 2.0 313M", "level": "variant", "family_id": "datadog-toto-2-0", "developer": [ "datadog" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 313000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-06-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Toto 2.0 size variant.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2605.20119" ], "unverified": [] }, { "id": "datadog-toto-2-0-4m", "name": "Toto 2.0 4M", "level": "variant", "family_id": "datadog-toto-2-0", "developer": [ "datadog" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 4000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-06-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Toto 2.0 size variant.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2605.20119" ], "unverified": [] }, { "id": "decart-lucy-2-0", "name": "Lucy 2.0", "level": "model", "family_id": null, "developer": [ "decart" ], "model_class": "video generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Decart video editing model (sibling of Oasis; also Lucy 1).", "verified_on": "2026-10-10", "sources": [ "https://dataconomy.com/2026/06/10/decart-lays-the-foundation-for-physical-ai-systems-with-oasis-3/" ], "unverified": [] }, { "id": "decart-oasis", "name": "Oasis", "level": "family", "family_id": null, "developer": [ "decart" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Decart real-time world model line.", "verified_on": "2026-10-10", "sources": [ "https://dataconomy.com/2026/06/10/decart-lays-the-foundation-for-physical-ai-systems-with-oasis-3/" ], "unverified": [] }, { "id": "decart-oasis-1", "name": "Oasis 1", "level": "model", "family_id": "decart-oasis", "developer": [ "decart" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-10", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First Oasis model.", "verified_on": "2026-10-10", "sources": [ "https://dataconomy.com/2026/06/10/decart-lays-the-foundation-for-physical-ai-systems-with-oasis-3/" ], "unverified": [] }, { "id": "decart-oasis-2", "name": "Oasis 2", "level": "model", "family_id": "decart-oasis", "developer": [ "decart" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-09", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier Oasis model.", "verified_on": "2026-10-10", "sources": [ "https://dataconomy.com/2026/06/10/decart-lays-the-foundation-for-physical-ai-systems-with-oasis-3/" ], "unverified": [] }, { "id": "decart-oasis-3", "name": "Oasis 3", "level": "model", "family_id": "decart-oasis", "developer": [ "decart" ], "model_class": "world model", "modalities_in": [ "action" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": "Real-time action-conditioned video; 22 fps at 768px; under 200 ms latency; 3 synchronised camera views; no duration limit", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-06", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Decart real-time world model served via a live API.", "verified_on": "2026-10-10", "sources": [ "https://dataconomy.com/2026/06/10/decart-lays-the-foundation-for-physical-ai-systems-with-oasis-3/", "https://pulse2.com/decart-300-million-raised-at-nearly-4-billion-valuation-raised-to-make-switching-ai-chips-easier/amp/" ], "unverified": [] }, { "id": "deemos-technology-rodin-gen-2", "name": "Rodin Gen-2", "level": "model", "family_id": null, "developer": [ "deemos-technology" ], "model_class": "3D", "modalities_in": [ "text", "image" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Deemos Hyper3D Rodin 3D generation model.", "verified_on": "2026-10-10", "sources": [ "https://www.yicaiglobal.com/news/chinese-ai-startup-deemos-raises-new-funds-for-hyper3d-rodins-global-rollout", "https://hyper3d.ai/" ], "unverified": [] }, { "id": "deepant", "name": "DeepAnT", "level": "model", "family_id": null, "developer": [], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "CNN-based anomaly detector on TSB-AD.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2602.01359" ], "unverified": [] }, { "id": "deepmodeling-dpa4", "name": "DPA4", "level": "model", "family_id": null, "developer": [ "deepmodeling" ], "model_class": "materials", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "GNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MLIP in a 2026 paper, not yet seen on the Matbench Discovery leaderboard.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2606.02419" ], "unverified": [] }, { "id": "deepseek-deepseek-coder-v2", "name": "DeepSeek-Coder-V2", "level": "model", "family_id": null, "developer": [ "deepseek" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "DeepSeekMoE; continued pre-training of a DeepSeek-V2 intermediate checkpoint with 6 trillion extra tokens", "params_total": 236000000000, "params_active": 21000000000, "params_checkpoint": null, "params_note": "Lite variants: 16B total, 2.4B active", "context_length": 128000, "release_date": "2024-06-14", "release_date_basis": "registry date", "weights_date": "2024-06-14", "openness": "open weights permissive", "licence": "deepseek-model-license", "access_routes": [ "API", "consumer app", "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Supports 338 programming languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "DeepSeek-Coder-V2 is an open mixture-of-experts code model (236B total, 21B active, 128K context) that DeepSeek reports as comparable to GPT-4 Turbo on code tasks.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/deepseek-ai/DeepSeek-Coder-V2-Instruct", "https://huggingface.co/api/models/deepseek-ai/DeepSeek-Coder-V2-Instruct" ], "unverified": [] }, { "id": "deepseek-deepseek-coder-v2-base", "name": "DeepSeek-Coder-V2-Base", "level": "model", "family_id": null, "developer": [ "deepseek" ], "model_class": "code", "modalities_in": [], "modalities_out": [], "architecture": "MoE", "architecture_detail": null, "params_total": 236000000000, "params_active": 21000000000, "params_checkpoint": null, "params_note": null, "context_length": 128000, "release_date": "2024-06-14", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "deepseek-model-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Base model of DeepSeek-Coder-V2-Instruct (236B total, 21B active).", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/deepseek-ai/DeepSeek-Coder-V2-Instruct", "https://huggingface.co/deepseek-ai/DeepSeek-Coder-V2-Base" ], "unverified": [] }, { "id": "deepseek-deepseek-llm-67b-base", "name": "deepseek-llm-67b-base", "level": "model", "family_id": null, "developer": [ "deepseek" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": 67000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-11-29", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Base model from which deepseek-llm-67b-chat was initialised.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/deepseek-ai/deepseek-llm-67b-chat", "https://huggingface.co/deepseek-ai/deepseek-llm-67b-base" ], "unverified": [] }, { "id": "deepseek-deepseek-llm-67b-chat", "name": "DeepSeek LLM 67B Chat (V1)", "level": "model", "family_id": null, "developer": [ "deepseek" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 67000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-11-29", "release_date_basis": "registry date", "weights_date": "2023-11-29", "openness": "open weights permissive", "licence": "deepseek-model-license", "access_routes": [ "open weights download", "consumer app" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Trained from scratch on 2 trillion tokens in English and Chinese", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "deepseek-llm-67b-chat is the first-generation DeepSeek 67B chat model, fine-tuned from deepseek-llm-67b-base on instruction data; commercial use is allowed under the model licence.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/deepseek-ai/deepseek-llm-67b-chat", "https://huggingface.co/api/models/deepseek-ai/deepseek-llm-67b-chat" ], "unverified": [] }, { "id": "deepseek-deepseek-ocr", "name": "DeepSeek-OCR", "level": "model", "family_id": null, "developer": [ "deepseek" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "autoregressive", "architecture_detail": null, "params_total": 3340000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10-17", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First DeepSeek OCR VLM (v1).", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2606.03264", "https://huggingface.co/api/models/deepseek-ai/DeepSeek-OCR" ], "unverified": [] }, { "id": "deepseek-deepseek-ocr-2", "name": "DeepSeek-OCR 2", "level": "model", "family_id": null, "developer": [ "deepseek" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "autoregressive", "architecture_detail": "VLM with optical token compression (unverified)", "params_total": 3390000000, "params_active": null, "params_checkpoint": null, "params_note": "3B in the research notes; HF reports 3.39B", "context_length": null, "release_date": "2026-01-27", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DeepSeek document OCR VLM, child of DeepSeek-OCR.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2606.03264", "https://huggingface.co/api/models/deepseek-ai/DeepSeek-OCR-2" ], "unverified": [] }, { "id": "deepseek-deepseek-r1", "name": "DeepSeek-R1", "level": "model", "family_id": null, "developer": [ "deepseek" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "RL stage cost from a secondary report of the Nature supplementary materials; GPU count not stated.", "context_length": null, "release_date": "2025-01", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DeepSeek reasoning model; teacher of the R1-Distill models via 800k generated samples. HF tree: 327 finetunes, 122 adapters, 70 quantizations, 5 merges.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B", "https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct", "https://eu.36kr.com/en/p/3471852231792260", "https://arxiv.org/abs/2501.12948" ], "unverified": [ "params_note" ] }, { "id": "deepseek-deepseek-r1-distill-llama-70b", "name": "DeepSeek-R1-Distill-Llama-70B", "level": "model", "family_id": null, "developer": [ "deepseek" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-01-20", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "llama-3-3-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DeepSeek-R1 distilled into a Llama base.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B", "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Llama-70B" ], "unverified": [] }, { "id": "deepseek-deepseek-r1-distill-llama-8b", "name": "DeepSeek-R1-Distill-Llama-8B", "level": "model", "family_id": null, "developer": [ "deepseek" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-01-20", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "llama-3-1-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DeepSeek-R1 distilled into a Llama base.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B", "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Llama-8B" ], "unverified": [] }, { "id": "deepseek-deepseek-r1-distill-qwen-1-5b", "name": "DeepSeek-R1-Distill-Qwen-1.5B", "level": "model", "family_id": null, "developer": [ "deepseek" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Licence: MIT (DeepSeek) plus Apache-2.0 upstream", "context_length": null, "release_date": "2025-01-20", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DeepSeek-R1 distilled into a Qwen base.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B", "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B" ], "unverified": [] }, { "id": "deepseek-deepseek-r1-distill-qwen-14b", "name": "DeepSeek-R1-Distill-Qwen-14B", "level": "model", "family_id": null, "developer": [ "deepseek" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Licence: MIT (DeepSeek) plus Apache-2.0 upstream", "context_length": null, "release_date": "2025-01-20", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DeepSeek-R1 distilled into a Qwen base.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B", "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-14B" ], "unverified": [] }, { "id": "deepseek-deepseek-r1-distill-qwen-32b", "name": "DeepSeek-R1-Distill-Qwen-32B", "level": "model", "family_id": null, "developer": [ "deepseek" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Licence: MIT (DeepSeek) plus Apache-2.0 upstream", "context_length": null, "release_date": "2025-01-20", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DeepSeek-R1 distilled into a Qwen base.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B" ], "unverified": [] }, { "id": "deepseek-deepseek-r1-distill-qwen-7b", "name": "DeepSeek-R1-Distill-Qwen-7B", "level": "model", "family_id": null, "developer": [ "deepseek" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Licence: MIT (DeepSeek) plus Apache-2.0 upstream", "context_length": null, "release_date": "2025-01-20", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DeepSeek-R1 distilled into a Qwen base.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B", "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-7B" ], "unverified": [] }, { "id": "deepseek-deepseek-v2", "name": "DeepSeek-V2", "level": "family", "family_id": null, "developer": [ "deepseek" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-05-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "DeepSeek-V2 is the DeepSeek series of Mixture-of-Experts language models, including V2.5.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/deepseek-ai/DeepSeek-V2.5-1210", "https://arxiv.org/abs/2405.04434" ], "unverified": [] }, { "id": "deepseek-deepseek-v2-5", "name": "DeepSeek-V2.5", "level": "model", "family_id": "deepseek-deepseek-v2", "developer": [ "deepseek" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-09-05", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "deepseek-model-license", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "DeepSeek V2.5 model, the predecessor that DeepSeek-V2.5-1210 upgrades.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/deepseek-ai/DeepSeek-V2.5-1210", "https://huggingface.co/deepseek-ai/DeepSeek-V2.5" ], "unverified": [] }, { "id": "deepseek-deepseek-v2-5-1210", "name": "DeepSeek-V2.5 (Dec 2024, DeepSeek-V2.5-1210)", "level": "model", "family_id": "deepseek-deepseek-v2", "developer": [ "deepseek" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 236000000000, "params_active": 21000000000, "params_checkpoint": null, "params_note": "Active count from Artificial Analysis", "context_length": 128000, "release_date": "2024-12-10", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights restricted", "licence": "deepseek-model-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "Upgraded DeepSeek-V2.5 MoE model with gains in math, coding, writing and reasoning; Artificial Analysis lists it as deprecated.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/deepseek-ai/DeepSeek-V2.5-1210", "https://artificialanalysis.ai/models/deepseek-v2-5", "https://github.com/deepseek-ai/DeepSeek-V2/blob/main/LICENSE-MODEL" ], "unverified": [] }, { "id": "deepseek-deepseek-v3", "name": "DeepSeek v3", "level": "model", "family_id": null, "developer": [ "deepseek" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 671000000000, "params_active": 37000000000, "params_checkpoint": null, "params_note": "Trained on 14.8T tokens; final run 2.788M H800 GPU hours, $5.576M at $2 per GPU hour, excluding prior research and ablations.", "context_length": null, "release_date": "2024-12", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Model used as an MLPerf Training v6.0 workload.", "verified_on": "2026-10-10", "sources": [ "https://mlcommons.org/benchmarks/training/", "https://arxiv.org/html/2412.19437v1" ], "unverified": [] }, { "id": "deepseek-deepseek-v3-1", "name": "DeepSeek-V3.1", "level": "model", "family_id": null, "developer": [ "deepseek" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "Hybrid thinking and non-thinking modes in one model; trained with UE8M0 FP8 scale format", "params_total": 671000000000, "params_active": 37000000000, "params_checkpoint": null, "params_note": null, "context_length": 128000, "release_date": "2025-08-21", "release_date_basis": "announcement page", "weights_date": "2025-08-21", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "API", "consumer app", "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DeepSeek hybrid model with thinking and non-thinking modes and stronger tool use and agent skills, post-trained on a long-context extension of the V3 base.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/deepseek-ai/DeepSeek-V3.1", "https://api-docs.deepseek.com/news/news250821", "https://openrouter.ai/deepseek/deepseek-chat-v3.1" ], "unverified": [] }, { "id": "deepseek-deepseek-v3-1-terminus", "name": "DeepSeek-V3.1-Terminus", "level": "model", "family_id": null, "developer": [ "deepseek" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-09-22", "release_date_basis": "registry date", "weights_date": "2025-09-22", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "API", "consumer app", "open weights download" ], "deployment_tiers": [], "footprint_note": "self_attn.o_proj parameters do not conform to the UE8M0 FP8 scale format in this checkpoint (known issue per model card)", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Update of DeepSeek-V3.1 with better language consistency and stronger code and search agents.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/deepseek-ai/DeepSeek-V3.1-Terminus", "https://api-docs.deepseek.com/news/news250922", "https://openrouter.ai/deepseek/deepseek-v3.1-terminus" ], "unverified": [] }, { "id": "deepseek-deepseek-v3-2", "name": "DeepSeek-V3.2", "level": "model", "family_id": null, "developer": [ "deepseek" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "DeepSeek Sparse Attention (DSA) for long-context efficiency; thinking integrated into tool use", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-12-01", "release_date_basis": "announcement page", "weights_date": "2025-12-01", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "API", "consumer app", "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Reasoning-first DeepSeek model built for agents and the official successor to V3.2-Exp, the first DeepSeek model to integrate thinking into tool use.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/deepseek-ai/DeepSeek-V3.2", "https://api-docs.deepseek.com/news/news251201", "https://openrouter.ai/deepseek/deepseek-v3.2" ], "unverified": [] }, { "id": "deepseek-deepseek-v3-2-exp", "name": "DeepSeek-V3.2-Exp", "level": "model", "family_id": null, "developer": [ "deepseek" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": "DeepSeek sparse attention", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-09-29", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Experimental DeepSeek model with sparse attention, released with an API price cut of over 50%.", "verified_on": "2026-10-10", "sources": [ "https://api-docs.deepseek.com/news/news250929" ], "unverified": [] }, { "id": "deepseek-deepseek-v4", "name": "DeepSeek V4", "level": "family", "family_id": null, "developer": [ "deepseek" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04-22", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DeepSeek's V4 family of MIT-licensed MoE models with 1M context.", "verified_on": "2026-10-10", "sources": [ "https://rits.shanghai.nyu.edu/ai/deepseek-releases-v4-open-source-1-6t-moe-with-1m-context", "https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro", "https://huggingface.co/deepseek-ai", "https://github.com/NVIDIA/TensorRT-LLM" ], "unverified": [] }, { "id": "deepseek-deepseek-v4-1-flash", "name": "DeepSeek-V4.1-Flash", "level": "model", "family_id": "deepseek-deepseek-v4", "developer": [ "deepseek" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "MoE \"causal encoder-decoder\" (20 + 20 layers), 384 routed experts (6 active + 1 shared), Engram memory module (196B), CSA2 attention", "params_total": 552000000000, "params_active": 16000000000, "params_checkpoint": 763000000000, "params_note": "763B listed (552B backbone); 8B active prefill, 16B decode", "context_length": 1000000, "release_date": "2026-09-10", "release_date_basis": "secondary report", "weights_date": "2026-10-02", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "API", "open weights download" ], "deployment_tiers": [ "multi-GPU node" ], "footprint_note": "T3 estimate (unverified). HF updated about 2026-10-02.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DeepSeek open MoE model trained from scratch on 45T multimodal tokens.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash", "https://huggingface.co/deepseek-ai", "https://artificialanalysis.ai/leaderboards/models", "https://arena.ai/leaderboard/chat/text", "https://slash-digital.io/en/insights/ai-models-2026/", "https://codersera.com/blog/open-source-llms-landscape-2026/" ], "unverified": [ "deployment_tiers" ] }, { "id": "deepseek-deepseek-v4-flash", "name": "DeepSeek-V4-Flash", "level": "model", "family_id": "deepseek-deepseek-v4", "developer": [ "deepseek" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 284000000000, "params_active": 13000000000, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": "2026-04-24", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "multi-GPU node", "single datacentre GPU" ], "footprint_note": "T3 estimate (unverified). Output 384K.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Smaller DeepSeek V4 MoE model (284B / 13B active), text only, with a 0731 update; base model released separately.", "verified_on": "2026-10-10", "sources": [ "https://blog.buildfastwithai.com/deepseek-v4-pro-review-2026", "https://huggingface.co/deepseek-ai", "https://leaderboard.steel.dev/leaderboards/swe-bench-verified.md", "https://rits.shanghai.nyu.edu/ai/deepseek-releases-v4-open-source-1-6t-moe-with-1m-context", "https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro" ], "unverified": [ "deployment_tiers" ] }, { "id": "deepseek-deepseek-v4-flash-0731", "name": "DeepSeek-V4-Flash-0731", "level": "variant", "family_id": null, "developer": [ "deepseek" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-07-31", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/deepseek-ai", "https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-0731" ], "unverified": [] }, { "id": "deepseek-deepseek-v4-flash-base", "name": "DeepSeek-V4-Flash Base", "level": "variant", "family_id": "deepseek-deepseek-v4", "developer": [ "deepseek" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04-22", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Base model of DeepSeek-V4-Flash, released separately.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/deepseek-ai", "https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-Base" ], "unverified": [] }, { "id": "deepseek-deepseek-v4-flash-vision-exp", "name": "DeepSeek-V4-Flash Vision-Exp", "level": "variant", "family_id": "deepseek-deepseek-v4", "developer": [ "deepseek" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-09-01", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vision variant of DeepSeek-V4-Flash (HF 1 Sep).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/deepseek-ai" ], "unverified": [] }, { "id": "deepseek-deepseek-v4-pro", "name": "DeepSeek-V4-Pro", "level": "model", "family_id": "deepseek-deepseek-v4", "developer": [ "deepseek" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "Hybrid CSA + HCA attention, mHC, Muon optimiser; pretrained on over 32T tokens", "params_total": 1600000000000, "params_active": 49000000000, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": "2026-04-24", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "hyperscale cluster", "multi-GPU node" ], "footprint_note": "T2 estimate (unverified). Output 384K. Preview release.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DeepSeek's 1.6T-parameter MoE model released with MIT weights and code (no data). HF tree: finetunes 15, quants 27; 408k downloads/month.", "verified_on": "2026-10-10", "sources": [ "https://blog.buildfastwithai.com/deepseek-v4-pro-review-2026", "https://huggingface.co/deepseek-ai", "https://leaderboard.steel.dev/leaderboards/swe-bench-verified.md", "https://rits.shanghai.nyu.edu/ai/deepseek-releases-v4-open-source-1-6t-moe-with-1m-context", "https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro" ], "unverified": [ "deployment_tiers" ] }, { "id": "deepseek-deepseek-v4-pro-0813", "name": "DeepSeek-V4-Pro-0813", "level": "variant", "family_id": "deepseek-deepseek-v4", "developer": [ "deepseek" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-08-13", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Updated checkpoint of DeepSeek-V4-Pro (HF 13 Aug).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/deepseek-ai", "https://arena.ai/leaderboard/chat/text" ], "unverified": [] }, { "id": "delft-imaging-cad4tb", "name": "CAD4TB", "level": "model", "family_id": null, "developer": [ "delft-imaging" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Chest X-ray TB screening CAD software; developer attribution unverified in the notes.", "verified_on": "2026-10-10", "sources": [ "https://www.qure.ai/", "https://www.who.int/publications/i/item/9789240022676" ], "unverified": [ "developer", "indonesian_measured", "indonesia_hosting" ] }, { "id": "dkfz-nnu-net-v2", "name": "nnU-Net v2", "level": "model", "family_id": null, "developer": [ "dkfz", "helmholtz-imaging" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": "U-Net family segmentation framework with ResEnc presets", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Medical image segmentation framework from DKFZ and Helmholtz Imaging, Apache-2.0.", "verified_on": "2026-10-10", "sources": [ "https://github.com/MIC-DKFZ/nnUNet" ], "unverified": [] }, { "id": "dlrm-dcnv2", "name": "DLRM-dcnv2", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "recommendation", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Model used as an MLPerf Training v6.0 workload.", "verified_on": "2026-10-10", "sources": [ "https://mlcommons.org/benchmarks/training/", "https://arxiv.org/html/2008.13535" ], "unverified": [] }, { "id": "dmlc-xgboost", "name": "XGBoost", "level": "model", "family_id": null, "developer": [ "dmlc" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "gradient boosting", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-08-15", "release_date_basis": "registry date", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Latest 3.4.1 on PyPI 15 Aug 2026 (GitHub 3.4.0, 4 Aug)", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Gradient boosting library, still a practical standard for tabular data.", "verified_on": "2026-10-10", "sources": [ "https://github.com/dmlc/xgboost/releases", "https://pypi.org/pypi/xgboost/json", "https://github.com/dmlc/xgboost" ], "unverified": [] }, { "id": "drbaph-higgs-audio-v3-studio", "name": "Higgs-Audio-v3-Studio", "level": "model", "family_id": null, "developer": [ "drbaph" ], "model_class": "speech synthesis", "modalities_in": [ "text", "audio" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-07-02", "release_date_basis": "registry date", "weights_date": "2026-07-02", "openness": "open weights restricted", "licence": "boson-higgs-tts-3-research-non-commercial", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU", "offline capable" ], "footprint_note": "GGUF builds q4_k_m (8 GB VRAM) to bf16 (16 GB); Windows x64 with NVIDIA RTX 30xx, 40xx or 50xx and a CUDA 13 driver", "runtimes": [], "languages_note": "Card lists many languages including Indonesian, Javanese, Malay, Tagalog and Vietnamese", "indonesian_support": "listed by developer", "regional_languages": [ "Javanese" ], "indonesia_hosting": "unknown", "status": null, "summary": "Higgs-Audio-v3-Studio hosts GGUF builds and a native C++/CUDA engine for a Windows desktop app that runs Higgs Audio v3 locally for TTS, voice cloning and multi-speaker generation; it is not the upstream release.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/drbaph/Higgs-Audio-v3-Studio", "https://huggingface.co/api/models/drbaph/Higgs-Audio-v3-Studio" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "ds-cnn", "name": "DS-CNN", "level": "model", "family_id": null, "developer": [ "arm" ], "model_class": "TinyML", "modalities_in": [ "text" ], "modalities_out": [ "labels and boxes" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-11-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "microcontroller", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Model listed in Arm ML-Zoo, optimised for Arm IP (Cortex-A, Cortex-M, Mali, Ethos-U) in TensorFlow Lite format.", "verified_on": "2026-10-10", "sources": [ "https://github.com/ARM-software/ML-zoo", "https://arxiv.org/html/1711.07128", "https://arxiv.org/abs/1711.07128" ], "unverified": [] }, { "id": "e2-tts", "name": "E2 TTS", "level": "model", "family_id": null, "developer": [ "microsoft-corporation" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-06-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "TTS model listed as the genealogy of F5-TTS.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/SWivid/F5-TTS", "https://arxiv.org/html/2406.18009", "https://arxiv.org/abs/2406.18009" ], "unverified": [] }, { "id": "e5", "name": "E5", "level": "model", "family_id": null, "developer": [ "microsoft-corporation" ], "model_class": "embedding", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Embedding model on which SEA-LION-E5-Embedding is based; developer not stated in these sections.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/aisingapore", "https://arxiv.org/html/2212.03533" ], "unverified": [] }, { "id": "ecmwf-aifs", "name": "ECMWF AIFS (Single / ENS)", "level": "model", "family_id": null, "developer": [ "ecmwf" ], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "other", "architecture_detail": "Trained with open-source Anemoi framework; ENS 31 km", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [ "anemoi" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ECMWF AI forecasting system; AIFS Single operational end Feb 2025, AIFS ENS operational 1 Jul 2025, up to 20% gains over physics ENS on surface temperature.", "verified_on": "2026-10-10", "sources": [ "https://www.ecmwf.int/en/about/media-centre/news/2025/ecmwfs-ensemble-ai-forecasts-become-operational", "https://www.ecmwf.int/en/newsletter/187/news/implementation-aifs-v2" ], "unverified": [ "licence" ] }, { "id": "ecmwf-aifs-v2", "name": "AIFS v2", "level": "model", "family_id": null, "developer": [ "ecmwf" ], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "AIFS v2 implementation noted in ECMWF Newsletter 187.", "verified_on": "2026-10-10", "sources": [ "https://www.ecmwf.int/en/newsletter/187/news/implementation-aifs-v2" ], "unverified": [] }, { "id": "ecmwf-ifs-ens", "name": "IFS ENS", "level": "model", "family_id": null, "developer": [ "ecmwf" ], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Physics ensemble baseline on WeatherBench 2.", "verified_on": "2026-10-10", "sources": [ "https://researchsites.withgoogle.com/weatherbench/deterministic/" ], "unverified": [ "developer" ] }, { "id": "ecmwf-ifs-hres", "name": "IFS HRES", "level": "model", "family_id": null, "developer": [ "ecmwf" ], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Physics deterministic baseline on WeatherBench 2.", "verified_on": "2026-10-10", "sources": [ "https://researchsites.withgoogle.com/weatherbench/deterministic/" ], "unverified": [ "developer" ] }, { "id": "ecnu-qiyao-m", "name": "QiYao-M", "level": "model", "family_id": null, "developer": [ "ecnu", "huawei" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ECNU and Huawei pretrained forecasting model on GIFT-Eval.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/spaces/Salesforce/GIFT-Eval" ], "unverified": [] }, { "id": "elevenlabs-eleven", "name": "Eleven", "level": "family", "family_id": null, "developer": [ "elevenlabs" ], "model_class": "speech synthesis", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ElevenLabs text-to-speech models.", "verified_on": "2026-10-10", "sources": [ "https://elevenlabs.io/docs/models" ], "unverified": [] }, { "id": "elevenlabs-eleven-flash-v2-5", "name": "Eleven Flash v2.5", "level": "model", "family_id": "elevenlabs-eleven", "developer": [ "elevenlabs" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ElevenLabs low-latency TTS supporting Indonesian.", "verified_on": "2026-10-10", "sources": [ "https://elevenlabs.io/docs/models" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "elevenlabs-eleven-multilingual-v2", "name": "Eleven Multilingual v2", "level": "model", "family_id": "elevenlabs-eleven", "developer": [ "elevenlabs" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ElevenLabs multilingual TTS supporting Indonesian; resold by Pika.", "verified_on": "2026-10-10", "sources": [ "https://elevenlabs.io/docs/models", "https://pika.art/" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "elevenlabs-eleven-music", "name": "Eleven Music", "level": "model", "family_id": null, "developer": [ "elevenlabs" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-07", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English, Spanish, German, Japanese and more; Indonesian not listed", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ElevenLabs music model cleared for commercial use with labels and publishers; release Jul or Aug 2025 (sources conflict).", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/ElevenLabs", "https://elevenlabs.io/docs/capabilities/music" ], "unverified": [ "release_date" ] }, { "id": "elevenlabs-eleven-music-v2-5", "name": "Eleven Music v2.5", "level": "model", "family_id": null, "developer": [ "elevenlabs" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ElevenLabs music model named among recent flagship releases by Artificial Analysis.", "verified_on": "2026-10-10", "sources": [ "https://alphasignal.ai/news/artificial-analysis-music-arena-v1-1-crowns-suno-v6-across-1-000-harder-prompts", "https://digg.com/ai/g8egglmo" ], "unverified": [] }, { "id": "elevenlabs-eleven-turbo-v2-5", "name": "Eleven Turbo v2.5", "level": "model", "family_id": "elevenlabs-eleven", "developer": [ "elevenlabs" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "Deprecated ElevenLabs TTS supporting Indonesian; resold by Pika.", "verified_on": "2026-10-10", "sources": [ "https://elevenlabs.io/docs/models", "https://pika.art/" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "elevenlabs-eleven-v3", "name": "Eleven v3", "level": "model", "family_id": "elevenlabs-eleven", "developer": [ "elevenlabs" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-06", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "70+ languages", "indonesian_support": "listed by developer", "regional_languages": [ "Javanese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "ElevenLabs TTS model.", "verified_on": "2026-10-10", "sources": [ "https://elevenlabs.io/docs/models", "https://artificialanalysis.ai/text-to-speech/leaderboard" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "elevenlabs-eleven-v3-conversational", "name": "Eleven v3 Conversational", "level": "variant", "family_id": "elevenlabs-eleven-v3", "developer": [ "elevenlabs" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-02", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Conversational variant of Eleven v3.", "verified_on": "2026-10-10", "sources": [ "https://elevenlabs.io/docs/models" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "elevenlabs-eleven-v4", "name": "Eleven v4", "level": "model", "family_id": "elevenlabs-eleven", "developer": [ "elevenlabs" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-09", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "consumer app" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "90+ languages, 10,000 character limit", "indonesian_support": "listed by developer", "regional_languages": [ "Javanese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "ElevenLabs TTS model, 2nd on the Artificial Analysis TTS arena, with Indonesian and Javanese listed.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/text-to-speech/leaderboard", "https://elevenlabs.io/docs/models", "https://en.wikipedia.org/wiki/ElevenLabs" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "elevenlabs-eleven-v4-turbo", "name": "Eleven v4 Turbo", "level": "variant", "family_id": "elevenlabs-eleven-v4", "developer": [ "elevenlabs" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-09", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "90+ languages; median latency about 100 ms", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Low-latency Eleven v4, 1st on the Artificial Analysis TTS arena.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/text-to-speech/leaderboard", "https://elevenlabs.io/docs/models" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "emu3-5", "name": "Emu3.5", "level": "model", "family_id": null, "developer": [ "baai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "autoregressive", "architecture_detail": "Decoder-only transformer trained with next-token prediction, 64 layers; Discrete Diffusion Adaptation (DiDA) for parallel image token generation at inference", "params_total": 34100000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10-30", "release_date_basis": "inferred from page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Emu3.5, a 34B native multimodal model from BAAI that predicts next tokens across vision and language, including image generation.", "verified_on": "2026-10-11", "sources": [ "https://arxiv.org/abs/2510.26583" ], "unverified": [] }, { "id": "eosda-pysawit", "name": "PySawit (enhanced)", "level": "model", "family_id": null, "developer": [ "eosda" ], "model_class": "geospatial", "modalities_in": [], "modalities_out": [], "architecture": "other", "architecture_detail": "Biophysical crop model plus satellite vegetation indices and soil moisture, recalibrated", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "EOSDA oil palm yield model; vendor claims accuracy raised from 78% to up to 90% on an unnamed Indonesian estate.", "verified_on": "2026-10-10", "sources": [ "https://eos.com/blog/eosda-enhanced-pysawit-model-to-improve-its-accuracy/" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "epfl-pet-oam-xl", "name": "PET-OAM-XL", "level": "model", "family_id": null, "developer": [ "epfl" ], "model_class": "materials", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "GNN", "architecture_detail": "Equivariant GNN interatomic potential (MLIP)", "params_total": 730000000, "params_active": null, "params_checkpoint": null, "params_note": "Training data OMat24+sAlex+MPtrj", "context_length": null, "release_date": "2026-01-10", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Machine-learned interatomic potential on Matbench Discovery (rank 1 by CPS, date added 2026-01-10).", "verified_on": "2026-10-10", "sources": [ "https://matbench-discovery.materialsproject.org/", "https://actu.epfl.ch/news/a-new-reference-model-for-machine-learning-driven" ], "unverified": [ "developer" ] }, { "id": "equflash", "name": "EquFlash", "level": "model", "family_id": null, "developer": [], "model_class": "materials", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "GNN", "architecture_detail": "Equivariant GNN interatomic potential (MLIP)", "params_total": 28700000, "params_active": null, "params_checkpoint": null, "params_note": "Training data OMat24+MPtrj+sAlex", "context_length": null, "release_date": "2025-06-23", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Machine-learned interatomic potential on Matbench Discovery (rank 3 by CPS, date added 2025-06-23).", "verified_on": "2026-10-10", "sources": [ "https://matbench-discovery.materialsproject.org/" ], "unverified": [ "developer" ] }, { "id": "equiformerv3", "name": "EquiformerV3", "level": "model", "family_id": null, "developer": [ "massachusetts-institute-of-technology" ], "model_class": "materials", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "GNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Equivariant MLIP in a 2026 paper, not yet seen on the leaderboard.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2604.09130", "https://arxiv.org/html/2604.09130" ], "unverified": [] }, { "id": "eurollm-eurollm-22b", "name": "EuroLLM-22B (2512)", "level": "model", "family_id": null, "developer": [ "eurollm" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 22000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-07-31", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "EuroLLM consortium's 22B dense text model; card not read because of the rate limit.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/utter-project" ], "unverified": [ "licence", "context_length", "indonesian_support" ] }, { "id": "eurollm-eurollm-9b", "name": "EuroLLM-9B", "level": "model", "family_id": null, "developer": [ "eurollm" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 9000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-11-22", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/utter-project", "https://huggingface.co/utter-project/EuroLLM-9B" ], "unverified": [] }, { "id": "eurollm-euromoe-2-6b-a0-6b", "name": "EuroMoE-2.6B-A0.6B", "level": "model", "family_id": null, "developer": [ "eurollm" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 2600000000, "params_active": 600000000, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-09", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "laptop", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "EuroLLM's small MoE model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/utter-project" ], "unverified": [ "licence", "indonesian_support" ] }, { "id": "evolutionaryscale-esm-c", "name": "ESM C", "level": "model", "family_id": null, "developer": [ "evolutionaryscale" ], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "300M, 600M, 6B", "context_length": null, "release_date": "2024-12", "release_date_basis": "inferred from page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": "600M single GPU; 6B via API on biohub.ai with biosecurity guardrails", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "EvolutionaryScale protein language model; repo says available under the MIT licence.", "verified_on": "2026-10-10", "sources": [ "https://github.com/evolutionaryscale/esm" ], "unverified": [] }, { "id": "evolutionaryscale-esm3", "name": "ESM3", "level": "family", "family_id": null, "developer": [ "evolutionaryscale" ], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "other", "architecture_detail": "Multimodal generative protein LM", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-06-25", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Generative protein model debated as GPAI with systemic risk under the EU AI Act.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2605.01611", "https://www.evolutionaryscale.ai/blog/esm3-release", "https://huggingface.co/api/models/EvolutionaryScale/esm3-sm-open-v1" ], "unverified": [ "release_date" ] }, { "id": "evolutionaryscale-esm3-98b", "name": "ESM3 98B", "level": "model", "family_id": "evolutionaryscale-esm3", "developer": [ "evolutionaryscale" ], "model_class": "scientific", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": 98000000000, "params_active": null, "params_checkpoint": null, "params_note": "Trained with more than 1e24 FLOPs", "context_length": null, "release_date": "2024-06-25", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Closed 98B ESM3 protein model.", "verified_on": "2026-10-10", "sources": [ "https://www.evolutionaryscale.ai/blog/esm3-release" ], "unverified": [] }, { "id": "evolutionaryscale-esm3-open-1-4b", "name": "ESM3 open 1.4B", "level": "model", "family_id": "evolutionaryscale-esm3", "developer": [ "evolutionaryscale" ], "model_class": "scientific", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": 1400000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-06-25", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Open 1.4B ESM3 model released under MIT.", "verified_on": "2026-10-10", "sources": [ "https://www.evolutionaryscale.ai/blog/esm3-release" ], "unverified": [] }, { "id": "evolutionaryscale-esmfold2", "name": "ESMFold2", "level": "model", "family_id": null, "developer": [ "evolutionaryscale" ], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-05", "release_date_basis": "inferred from page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Structure model built on ESM C 6B (ID esmfold2-fast-2026-05).", "verified_on": "2026-10-10", "sources": [ "https://github.com/evolutionaryscale/esm" ], "unverified": [] }, { "id": "excarta-heal-vit", "name": "Excarta (HEAL-ViT)", "level": "model", "family_id": null, "developer": [], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-02-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Weather model on WeatherBench 2 (2020).", "verified_on": "2026-10-10", "sources": [ "https://researchsites.withgoogle.com/weatherbench/deterministic/", "https://arxiv.org/abs/2403.17016" ], "unverified": [] }, { "id": "featurespace-tallierltm", "name": "TallierLTM", "level": "model", "family_id": null, "developer": [ "featurespace" ], "model_class": "tabular", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": "Self-supervised Large Transaction Model with embedding API", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-10-27", "release_date_basis": "notes", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Featurespace fraud model; vendor claims up to 71% fraud value detection uplift when its embeddings are added. Acquisition by Visa unverified.", "verified_on": "2026-10-10", "sources": [ "https://paymentexpert.com/2023/10/27/featurespace-ai-large-transaction/" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "figure-ai-helix", "name": "Helix", "level": "model", "family_id": "figure-ai-helix-family", "developer": [ "figure-ai" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": "hybrid", "architecture_detail": "System 2: 7B open-weight VLM at 7 to 9 Hz; System 1: 80M visuomotor policy at 200 Hz; 35-DoF upper body", "params_total": 7080000000, "params_active": null, "params_checkpoint": null, "params_note": "7B + 80M", "context_length": null, "release_date": "2025-02-20", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "internal only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Figure AI VLA for its own robots, trained on about 500 h of teleoperation data.", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/helix_vla" ], "unverified": [] }, { "id": "figure-ai-helix-02", "name": "Helix 02", "level": "model", "family_id": "figure-ai-helix-family", "developer": [ "figure-ai" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": "hybrid", "architecture_detail": "Adds System 0: 10M-param whole-body controller at 1 kHz, trained in sim on 1,000+ h retargeted human motion", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "7B + 80M + 10M", "context_length": null, "release_date": "2026-01-27", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "internal only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Helix with whole-body control on Figure 03; Figure 02 loaded 90,000+ parts at BMW Spartanburg over 11 months.", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/helix_vla" ], "unverified": [] }, { "id": "figure-ai-helix-2-5", "name": "Helix 2.5", "level": "model", "family_id": "figure-ai-helix-family", "developer": [ "figure-ai" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": null, "architecture_detail": "Pretrained from random initialisation on Figure's Index human-video dataset", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-09-17", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "internal only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Helix version not built on a third-party VLM; vendor reports 56% whole-task completion in unseen homes.", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/helix_vla" ], "unverified": [] }, { "id": "figure-ai-helix-family", "name": "Helix", "level": "family", "family_id": null, "developer": [ "figure-ai" ], "model_class": "robot policy", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Figure AI humanoid VLA models.", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/helix_vla" ], "unverified": [] }, { "id": "fish-audio-fish-speech", "name": "Fish Speech", "level": "model", "family_id": null, "developer": [ "fish-audio" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier Fish Audio TTS line.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/fishaudio/s2-pro" ], "unverified": [] }, { "id": "fish-audio-s2-pro", "name": "Fish Audio S2 Pro", "level": "model", "family_id": null, "developer": [ "fish-audio" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "autoregressive", "architecture_detail": "Dual-AR decoder-only plus RVQ codec", "params_total": 5000000000, "params_active": null, "params_checkpoint": null, "params_note": "About 5B", "context_length": null, "release_date": "2026-03", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "fish-audio-research", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": "80+ languages including Indonesian (id)", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fish Audio open TTS model; commercial use needs a licence.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/fishaudio/s2-pro", "https://pulse2.com/fish-audio-raises-52-million-seed-round-to-expand-expressive-voice-ai-platform/" ], "unverified": [ "deployment_tiers", "indonesian_measured", "indonesia_hosting" ] }, { "id": "freedomintelligence-huatuogpt-o1-72b", "name": "HuatuoGPT-o1-72B", "level": "model", "family_id": null, "developer": [ "freedomintelligence" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 73000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-12-25", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Chinese medical reasoning LLM fine-tuned from Qwen2.5-72B-Instruct (arXiv 2412.18925); the model card lists no benchmarks. CUHK-Shenzhen affiliation unverified.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/FreedomIntelligence/HuatuoGPT-o1-72B" ], "unverified": [] }, { "id": "fudan-university-fuxi", "name": "FuXi", "level": "model", "family_id": null, "developer": [ "fudan-university" ], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "dense transformer", "architecture_detail": "Cascade machine learning forecasting system for 15-day global forecasts", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-06-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fudan University cascade weather model targeting the ECMWF ensemble mean at long leads.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2306.12873", "https://news.tuoitre.vn/china-bets-on-ai-weather-forecasting-as-extreme-weather-intensifies-103260812213327362.htm", "https://infoplaza.com/en/blog/ai-weather-forecasting-second-generation-ai-weather-models" ], "unverified": [] }, { "id": "fudan-university-fuxi-2-1", "name": "FuXi-2.1", "level": "model", "family_id": null, "developer": [ "fudan-university" ], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-06-18", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "FuXi weights on HF tpys/fuxi-2.1.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/api/models/tpys/fuxi-2.1", "https://github.com/tpys/FuXi" ], "unverified": [] }, { "id": "funaudiollm-cosyvoice-1", "name": "CosyVoice 1", "level": "model", "family_id": null, "developer": [ "funaudiollm" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First CosyVoice model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/FunAudioLLM/Fun-CosyVoice3-0.5B-2512" ], "unverified": [] }, { "id": "funaudiollm-cosyvoice-2", "name": "CosyVoice 2", "level": "model", "family_id": null, "developer": [ "funaudiollm" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier CosyVoice model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/FunAudioLLM/Fun-CosyVoice3-0.5B-2512" ], "unverified": [] }, { "id": "funaudiollm-fun-cosyvoice3-0-5b-2512", "name": "Fun-CosyVoice3-0.5B-2512", "level": "model", "family_id": null, "developer": [ "funaudiollm" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "hybrid", "architecture_detail": "LLM-based TTS plus flow matching", "params_total": 500000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": "9 languages plus 18 Chinese dialects; Indonesian not listed", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "FunAudioLLM open TTS model (CosyVoice 3).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/FunAudioLLM/Fun-CosyVoice3-0.5B-2512" ], "unverified": [] }, { "id": "g42-jais", "name": "Jais", "level": "model", "family_id": null, "developer": [ "g42" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Not on AA.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/leaderboards/models" ], "unverified": [ "status" ] }, { "id": "galbot-astrabrain", "name": "AstraBrain", "level": "model", "family_id": null, "developer": [ "galbot" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-02-16", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [ "internal only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Galbot end-to-end brain-cerebellum-neural control whole-body model shown at the Spring Festival Gala.", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/galbot" ], "unverified": [ "openness" ] }, { "id": "galbot-graspvla", "name": "GraspVLA", "level": "model", "family_id": null, "developer": [ "galbot", "peking-university" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": null, "architecture_detail": "Grasping VLA pretrained on SynGrasp-1B billion-frame synthetic data", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-05-06", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open grasping VLA from Galbot and Peking University (repo PKU-EPIC/GraspVLA, licence unverified).", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/galbot" ], "unverified": [ "licence" ] }, { "id": "galbot-groceryvla", "name": "GroceryVLA", "level": "model", "family_id": null, "developer": [ "galbot" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [ "internal only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Galbot retail manipulation VLA on Galbot robots.", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/galbot" ], "unverified": [ "openness" ] }, { "id": "galbot-trackvla", "name": "TrackVLA", "level": "model", "family_id": null, "developer": [ "galbot" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [ "internal only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Galbot navigation and tracking VLA.", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/galbot" ], "unverified": [ "openness" ] }, { "id": "google-amie", "name": "AMIE", "level": "model", "family_id": null, "developer": [ "google", "google-deepmind" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "video" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Multi-agent research system built on Gemini and Project Astra", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "research only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": "Google research system for diagnostic dialogue, extended to video consultations on 11 Aug 2026 and tested in a randomized study with simulated patients. Not clinically deployed.", "verified_on": "2026-10-10", "sources": [ "https://blog.google/innovation-and-ai/models-and-research/google-research/amie-video-consultations/" ], "unverified": [] }, { "id": "google-aml-ai", "name": "Anti Money Laundering AI (AML AI)", "level": "model", "family_id": null, "developer": [ "google", "hsbc" ], "model_class": "tabular", "modalities_in": [], "modalities_out": [], "architecture": "other", "architecture_detail": "Proprietary ML risk scoring", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "Managed cloud service", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Google Cloud AML transaction monitoring model co-developed with HSBC; HSBC reports 2 to 4x suspicious activity found and 60% fewer alerts.", "verified_on": "2026-10-10", "sources": [ "https://cloud.google.com/blog/topics/financial-services/how-hsbc-fights-money-launderers-with-artificial-intelligence", "https://services.google.com/fh/files/misc/hsbc_celent_model_risk_manager_of_the_year_2023.pdf" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "google-deepmind-alphaearth-foundations", "name": "AlphaEarth Foundations (Satellite Embedding v1 Annual)", "level": "model", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "other", "architecture_detail": "Embedding field model; 64-dim embedding per 10 m pixel", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-07", "release_date_basis": "inferred from page", "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [ "cloud marketplace" ], "deployment_tiers": [], "footprint_note": "Consumed via Earth Engine (cloud); embeddings published for 2017 to 2024", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google DeepMind geospatial embedding model; about 23.9% lower average error than next-best across 15 mapping tasks (vs CCDC, SatCLIP, Prithvi, Clay), ET R2 0.58.", "verified_on": "2026-10-10", "sources": [ "https://arxiviq.substack.com/p/alphaearth-foundations-an-embedding" ], "unverified": [ "openness" ] }, { "id": "google-deepmind-alphafold-2", "name": "AlphaFold 2", "level": "model", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-11", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Predecessor of AlphaFold 3.", "verified_on": "2026-10-10", "sources": [ "https://scalingbiotech.substack.com/p/open-source-protein-models-openfold" ], "unverified": [] }, { "id": "google-deepmind-alphafold-3", "name": "AlphaFold 3", "level": "model", "family_id": null, "developer": [ "google-deepmind", "isomorphic-labs" ], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "diffusion", "architecture_detail": "Diffusion over atom coordinates", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-05-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "alphafold-3-weights-terms", "access_routes": [ "open weights download", "API", "research only" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": "Code Apache-2.0; weights only if received directly from Google under separate terms", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Biomolecular structure model. At CASP16 the AF3 server was comparable to, not ahead of, the best groups (B8); AF3 was the core engine of most top groups.", "verified_on": "2026-10-10", "sources": [ "https://scalingbiotech.substack.com/p/open-source-protein-models-openfold", "https://github.com/google-deepmind/alphafold3", "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12750027/", "https://www.nature.com/articles/s41586-024-07487-w", "https://www.biopharmatrend.com/news/alphafold3-goes-open-source-1017/", "https://blog.google/technology/ai/google-deepmind-isomorphic-alphafold-3-ai-model/" ], "unverified": [] }, { "id": "google-deepmind-alphagenome", "name": "AlphaGenome", "level": "model", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "scientific", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Genomics model attributed to DeepMind; unverified (page not read).", "verified_on": "2026-10-10", "sources": [ "https://github.com/ArcInstitute/evo2" ], "unverified": [ "developer", "summary" ] }, { "id": "google-deepmind-antigravity-preview-09-2026", "name": "antigravity-preview-09-2026", "level": "model", "family_id": "google-deepmind-gemini", "developer": [ "google-deepmind" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": "Managed agent running in a Linux sandbox, offered in the Gemini API.", "verified_on": "2026-10-10", "sources": [ "https://ai.google.dev/gemini-api/docs/models" ], "unverified": [] }, { "id": "google-deepmind-deep-research-max-preview-04-2026", "name": "deep-research-max-preview-04-2026", "level": "model", "family_id": "google-deepmind-gemini", "developer": [ "google-deepmind" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": "Max tier of the deep research agent preview in the Gemini API.", "verified_on": "2026-10-10", "sources": [ "https://ai.google.dev/gemini-api/docs/models" ], "unverified": [] }, { "id": "google-deepmind-deep-research-preview-04-2026", "name": "deep-research-preview-04-2026", "level": "model", "family_id": "google-deepmind-gemini", "developer": [ "google-deepmind" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": "Deep research agent preview in the Gemini API.", "verified_on": "2026-10-10", "sources": [ "https://ai.google.dev/gemini-api/docs/models" ], "unverified": [] }, { "id": "google-deepmind-dreamer", "name": "Dreamer", "level": "family", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Model-based RL world-model agents by Hafner et al.", "verified_on": "2026-10-10", "sources": [ "https://www.nature.com/articles/s41586-025-08744-2", "https://danijar.com/project/dreamer4/" ], "unverified": [] }, { "id": "google-deepmind-dreamer-4", "name": "Dreamer 4", "level": "model", "family_id": "google-deepmind-dreamer", "developer": [ "google-deepmind" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": "other", "architecture_detail": "New objective and architecture; real-time interactive inference on a single GPU", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-09", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "research only" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First agent to obtain Minecraft diamonds purely from offline data; beats OpenAI VPT offline with 100x less data.", "verified_on": "2026-10-10", "sources": [ "https://danijar.com/project/dreamer4/" ], "unverified": [ "developer" ] }, { "id": "google-deepmind-dreamerv3", "name": "DreamerV3", "level": "model", "family_id": "google-deepmind-dreamer", "developer": [ "google-deepmind" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": "other", "architecture_detail": "Learned world model plus imagination-based actor-critic; single configuration", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-04-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "research only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Model-based RL agent (Hafner et al., Nature 2025) that beats specialised methods on 150+ tasks and first collected Minecraft diamonds from scratch.", "verified_on": "2026-10-10", "sources": [ "https://www.nature.com/articles/s41586-025-08744-2" ], "unverified": [ "developer" ] }, { "id": "google-deepmind-gemini", "name": "Gemini", "level": "family", "family_id": null, "developer": [ "google-deepmind", "google" ], "model_class": "multimodal LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google DeepMind's proprietary model family, from Gemini 2.5 through Gemini 3.x Flash to Gemini 4 Argon.", "verified_on": "2026-10-10", "sources": [ "https://ai.google.dev/gemini-api/docs/models", "https://blog.google/innovation-and-ai/models-and-research/google-research/amie-video-consultations/", "https://www.designnews.com/automation/honeywell-uses-google-cloud-ai-to-accelerate-autonomous-operations", "https://ai.google.dev/gemini-api/docs/models/gemini-3.1-pro-preview", "https://ai.google.dev/gemini-api/docs/models/gemini-2.5-flash" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "google-deepmind-gemini-2-0", "name": "Gemini 2.0", "level": "model", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "multimodal LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Gemini model that Gemini Robotics is built on.", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/gemini_robotics_2" ], "unverified": [] }, { "id": "google-deepmind-gemini-2-5-computer-use-preview-10-2025", "name": "gemini-2.5-computer-use-preview-10-2025", "level": "model", "family_id": "google-deepmind-gemini", "developer": [ "google-deepmind" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "shut down", "summary": "Computer-use preview model, shut down.", "verified_on": "2026-10-10", "sources": [ "https://ai.google.dev/gemini-api/docs/models" ], "unverified": [] }, { "id": "google-deepmind-gemini-2-5-flash", "name": "Gemini 2.5 Flash", "level": "model", "family_id": "google-deepmind-gemini", "developer": [ "google-deepmind", "google" ], "model_class": "multimodal LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "Gemini 2.5 model, limited to past users.", "verified_on": "2026-10-10", "sources": [ "https://ai.google.dev/gemini-api/docs/models", "https://cognite.com/en/atlas-ai-slm-llm-benchmark-report", "https://ai.google.dev/gemini-api/docs/models/gemini-2.5-flash" ], "unverified": [ "indonesian_measured" ] }, { "id": "google-deepmind-gemini-2-5-flash-lite", "name": "Gemini 2.5 Flash-Lite", "level": "model", "family_id": "google-deepmind-gemini", "developer": [ "google-deepmind" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "Gemini 2.5 model, limited to past users.", "verified_on": "2026-10-10", "sources": [ "https://ai.google.dev/gemini-api/docs/models" ], "unverified": [] }, { "id": "google-deepmind-gemini-2-5-pro", "name": "Gemini 2.5 Pro", "level": "model", "family_id": "google-deepmind-gemini", "developer": [ "google-deepmind" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "Gemini 2.5 model, limited to past users.", "verified_on": "2026-10-10", "sources": [ "https://ai.google.dev/gemini-api/docs/models" ], "unverified": [] }, { "id": "google-deepmind-gemini-3-1-flash-lite", "name": "Gemini 3.1 Flash-Lite", "level": "model", "family_id": "google-deepmind-gemini", "developer": [ "google-deepmind" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://ai.google.dev/gemini-api/docs/models" ], "unverified": [] }, { "id": "google-deepmind-gemini-3-1-pro", "name": "Gemini 3.1 Pro (preview)", "level": "model", "family_id": "google-deepmind-gemini", "developer": [ "google-deepmind", "google" ], "model_class": "multimodal LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": "2026-02-19", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "Model ID gemini-3.1-pro-preview; newest Pro on the Gemini API.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": "Preview Pro-tier Gemini model, still the newest Pro on the Gemini API.", "verified_on": "2026-10-10", "sources": [ "https://ai.google.dev/gemini-api/docs/models", "https://arena.ai/leaderboard/chat/text", "https://leaderboard.steel.dev/leaderboards/swe-bench-verified.md", "https://labs.scale.com/leaderboard/swe_bench_pro_public", "https://www.datalearner.com/en/leaderboards/external/aa-quality-index", "https://ai.google.dev/gemini-api/docs/models/gemini-3.1-pro-preview" ], "unverified": [] }, { "id": "google-deepmind-gemini-3-5-flash", "name": "Gemini 3.5 Flash", "level": "model", "family_id": "google-deepmind-gemini", "developer": [ "google-deepmind" ], "model_class": "multimodal LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-05", "release_date_basis": "notes", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Gemini Flash model of May 2026; retired from Copilot 2026-10-02.", "verified_on": "2026-10-10", "sources": [ "https://ai.google.dev/gemini-api/docs/models", "https://noqta.tn/en/blog/gemini-3-6-flash-antigravity-agentic-coding-2026", "https://docs.github.com/en/copilot/reference/ai-models/supported-models", "https://aiwiki.ai/wiki/gemini_robotics_2", "https://ai.google.dev/gemini-api/docs/models/gemini-3.5-flash" ], "unverified": [] }, { "id": "google-deepmind-gemini-3-5-pro", "name": "Gemini 3.5 Pro", "level": "model", "family_id": "google-deepmind-gemini", "developer": [ "google-deepmind" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "internal only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "announced", "summary": "Reportedly delayed and not released; held back for partners (secondary reports).", "verified_on": "2026-10-10", "sources": [ "https://ai.google.dev/gemini-api/docs/models", "https://noqta.tn/en/blog/gemini-3-6-flash-antigravity-agentic-coding-2026" ], "unverified": [ "status" ] }, { "id": "google-deepmind-gemini-3-6-flash", "name": "Gemini 3.6 Flash", "level": "model", "family_id": "google-deepmind-gemini", "developer": [ "google-deepmind" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": "2026-07-21", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "1M+ context, 65,536 output tokens. Retired from Copilot 2026-10-02.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Gemini Flash model of July 2026 used for agentic coding.", "verified_on": "2026-10-10", "sources": [ "https://noqta.tn/en/blog/gemini-3-6-flash-antigravity-agentic-coding-2026", "https://docs.github.com/en/copilot/reference/ai-models/supported-models" ], "unverified": [] }, { "id": "google-deepmind-gemini-3-7-flash", "name": "Gemini 3.7 Flash", "level": "model", "family_id": "google-deepmind-gemini", "developer": [ "google-deepmind" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-08", "release_date_basis": "notes", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Gemini Flash model of August 2026; GA in GitHub Copilot.", "verified_on": "2026-10-10", "sources": [ "https://ai.google.dev/gemini-api/docs/models", "https://arena.ai/leaderboard/chat/vision", "https://docs.github.com/en/copilot/reference/ai-models/supported-models" ], "unverified": [] }, { "id": "google-deepmind-gemini-3-8-flash", "name": "Gemini 3.8 Flash", "level": "model", "family_id": "google-deepmind-gemini", "developer": [ "google-deepmind" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1048576, "release_date": "2026-09-02", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "GA in Gemini API as gemini-3.8-flash (New, Stable). Output 64K. Also in Antigravity and GitHub Copilot.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Gemini Flash model built for long-horizon software engineering, GA in the Gemini API. Modalities not stated on the page read.", "verified_on": "2026-10-10", "sources": [ "https://ai.google.dev/gemini-api/docs/models", "https://slash-digital.io/en/insights/ai-models-2026/", "https://artificialanalysis.ai/leaderboards/models", "https://arena.ai/leaderboard/chat/text", "https://ai.google.dev/gemini-api/docs/models", "https://deepmind.google/models/gemini/flash/", "https://docs.github.com/en/copilot/reference/ai-models/supported-models" ], "unverified": [ "modalities_in" ] }, { "id": "google-deepmind-gemini-3-8-live", "name": "gemini-3.8-live", "level": "model", "family_id": "google-deepmind-gemini", "developer": [ "google-deepmind" ], "model_class": "speech synthesis", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Audio-to-audio voice agent model in the Gemini API.", "verified_on": "2026-10-10", "sources": [ "https://ai.google.dev/gemini-api/docs/models" ], "unverified": [] }, { "id": "google-deepmind-gemini-3-8-live-extended-thinking", "name": "gemini-3.8-live-extended-thinking", "level": "model", "family_id": "google-deepmind-gemini", "developer": [ "google-deepmind" ], "model_class": "speech synthesis", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Extended-thinking variant of the Gemini 3.8 Live voice agent model.", "verified_on": "2026-10-10", "sources": [ "https://ai.google.dev/gemini-api/docs/models" ], "unverified": [] }, { "id": "google-deepmind-gemini-3-deep-think", "name": "Gemini 3 Deep Think", "level": "model", "family_id": "google-deepmind-gemini", "developer": [ "google-deepmind" ], "model_class": "reasoning", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-11-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app", "gated preview" ], "deployment_tiers": [], "footprint_note": "Specialized reasoning mode of Gemini 3; Google AI Ultra subscribers in the Gemini app, early access via Gemini API by expression of interest. Major upgrade on 12 February 2026", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Enhanced reasoning mode of Gemini 3 for hard problems in science, research and engineering.", "verified_on": "2026-10-11", "sources": [ "https://blog.google/products/gemini/gemini-3/", "https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-deep-think/" ], "unverified": [] }, { "id": "google-deepmind-gemini-3-pro", "name": "Gemini 3 Pro", "level": "model", "family_id": "google-deepmind-gemini", "developer": [ "google-deepmind", "google" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025", "release_date_basis": "notes", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "shut down", "summary": "Gemini 3 Pro preview, shut down.", "verified_on": "2026-10-10", "sources": [ "https://ai.google.dev/gemini-api/docs/models", "https://arxiv.org/pdf/2606.03264", "https://www.benchlm.ai/benchmarks/mmmu", "https://ai.google.dev/gemini-api/docs/models/gemini-3-pro-preview", "https://ai.google.dev/gemini-api/docs/deprecations" ], "unverified": [] }, { "id": "google-deepmind-gemini-4-argon", "name": "Gemini 4 Argon", "level": "model", "family_id": "google-deepmind-gemini", "developer": [ "google-deepmind" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": "2026-09-30", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "gated preview" ], "deployment_tiers": [], "footprint_note": "Gated: Fairwind Program cyberdefenders first, then paid API and AI Ultra; no public model ID and not in the Gemini API model list on 2026-10-10. Output limit raised to 1M per secondary report.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": "Google DeepMind's newest frontier Gemini model, ranked first on Arena Text; modalities not specified. llm-stats labels it lightweight (unverified).", "verified_on": "2026-10-10", "sources": [ "https://www.orcarouter.ai/es/blog/gemini-4-argon", "https://ai.google.dev/gemini-api/docs/models", "https://artificialanalysis.ai/leaderboards/models", "https://arena.ai/leaderboard/chat/text", "https://www.digitalapplied.com/blog/ai-model-releases-october-2026-tracker", "https://arena.ai/leaderboard/text" ], "unverified": [] }, { "id": "google-deepmind-gemini-omni-flash", "name": "Gemini Omni Flash", "level": "model", "family_id": "google-gemini-omni", "developer": [ "google-deepmind", "google" ], "model_class": "video generation", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "video", "audio" ], "architecture": null, "architecture_detail": "Unified pipeline combining Gemini, Nano Banana, Veo", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-05-19", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google media model, noted in passing in the research notes.", "verified_on": "2026-10-10", "sources": [ "https://ai.google.dev/gemini-api/docs/models", "https://framia.converge.ai/page/en-US/news/gemini-omni-ai-video-generation-model", "https://arena.ai/leaderboard/text-to-video" ], "unverified": [] }, { "id": "google-deepmind-gemini-robotics", "name": "Gemini Robotics", "level": "model", "family_id": "google-deepmind-gemini-robotics-family", "developer": [ "google-deepmind" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03-12", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "gated preview" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Gemini 2.0-based VLA for trusted testers; Apptronik humanoid partner.", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/gemini_robotics_2" ], "unverified": [] }, { "id": "google-deepmind-gemini-robotics-1-5", "name": "Gemini Robotics 1.5", "level": "model", "family_id": "google-deepmind-gemini-robotics-family", "developer": [ "google-deepmind" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-09-25", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "gated preview" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Agentic VLA that thinks in language before acting, with cross-embodiment motion transfer.", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/gemini_robotics_2" ], "unverified": [] }, { "id": "google-deepmind-gemini-robotics-2", "name": "Gemini Robotics 2", "level": "family", "family_id": "google-deepmind-gemini-robotics-family", "developer": [ "google-deepmind" ], "model_class": "robot policy", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-07-30", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Gemini Robotics 2 family (VLA, ER 2, On-Device 2); partners Apptronik, Boston Dynamics, Agile Robots.", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/gemini_robotics_2" ], "unverified": [] }, { "id": "google-deepmind-gemini-robotics-2-vla", "name": "Gemini Robotics 2 VLA", "level": "model", "family_id": "google-deepmind-gemini-robotics-2", "developer": [ "google-deepmind" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": null, "architecture_detail": "Single learned policy for a whole humanoid (feet to fingertips)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-07-30", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "gated preview" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Whole-body humanoid VLA in private preview for trusted testers.", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/gemini_robotics_2" ], "unverified": [] }, { "id": "google-deepmind-gemini-robotics-er", "name": "Gemini Robotics-ER", "level": "model", "family_id": "google-deepmind-gemini-robotics-family", "developer": [ "google-deepmind" ], "model_class": "multimodal LLM", "modalities_in": [ "image", "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03-12", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "gated preview" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Embodied-reasoning VLM.", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/gemini_robotics_2" ], "unverified": [] }, { "id": "google-deepmind-gemini-robotics-er-1-5", "name": "Gemini Robotics-ER 1.5", "level": "model", "family_id": "google-deepmind-gemini-robotics-family", "developer": [ "google-deepmind" ], "model_class": "multimodal LLM", "modalities_in": [ "image", "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-09-25", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Embodied-reasoning model in the Gemini API.", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/gemini_robotics_2" ], "unverified": [] }, { "id": "google-deepmind-gemini-robotics-er-1-6", "name": "Gemini Robotics-ER 1.6", "level": "model", "family_id": "google-deepmind-gemini-robotics-family", "developer": [ "google-deepmind", "boston-dynamics" ], "model_class": "multimodal LLM", "modalities_in": [ "image", "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04-14", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Embodied reasoning VLM with instrument reading and multi-view, used for Spot inspection with Boston Dynamics.", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/gemini_robotics_2" ], "unverified": [] }, { "id": "google-deepmind-gemini-robotics-er-2", "name": "Gemini Robotics-ER 2", "level": "model", "family_id": "google-deepmind-gemini-robotics-2", "developer": [ "google-deepmind" ], "model_class": "multimodal LLM", "modalities_in": [ "image", "video", "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Orchestrates the VLA as a tool with bidirectional streaming; built on Gemini 3.5 Flash", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-07-30", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": "Embodied reasoning model in public preview in AI Studio and the Gemini API.", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/gemini_robotics_2" ], "unverified": [] }, { "id": "google-deepmind-gemini-robotics-er-2-preview", "name": "gemini-robotics-er-2-preview", "level": "model", "family_id": "google-deepmind-gemini", "developer": [ "google-deepmind" ], "model_class": "multimodal LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": "Gemini robotics model in preview in the Gemini API.", "verified_on": "2026-10-10", "sources": [ "https://ai.google.dev/gemini-api/docs/models" ], "unverified": [] }, { "id": "google-deepmind-gemini-robotics-family", "name": "Gemini Robotics", "level": "family", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "robot policy", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google DeepMind Gemini-based robotics models.", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/gemini_robotics_2" ], "unverified": [] }, { "id": "google-deepmind-gemini-robotics-on-device", "name": "Gemini Robotics On-Device", "level": "model", "family_id": "google-deepmind-gemini-robotics-family", "developer": [ "google-deepmind" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-06-24", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "gated preview" ], "deployment_tiers": [], "footprint_note": "Runs locally on robot; SDK with MuJoCo", "runtimes": [ "mujoco" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Local on-robot VLA, the first Gemini Robotics model offered for fine-tuning.", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/gemini_robotics_2" ], "unverified": [] }, { "id": "google-deepmind-gemini-robotics-on-device-2", "name": "Gemini Robotics On-Device 2", "level": "model", "family_id": "google-deepmind-gemini-robotics-2", "developer": [ "google-deepmind" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": null, "architecture_detail": "Outputs numeric actions; built on Gemma plus Gemini Robotics 1.5 technology", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-07-30", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "gated preview" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "On-device robotics model for trusted testers.", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/gemini_robotics_2" ], "unverified": [] }, { "id": "google-deepmind-gencast", "name": "GenCast (WeatherNext 1 Gen)", "level": "model", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "diffusion", "architecture_detail": "Diffusion ensemble on a mesh; 0.25 deg / 1 deg; Mini 1 deg", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-sa-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": "Mini runs on Colab", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Ensemble weather model; beat ECMWF ENS on more than 96% of targets; code Apache-2.0.", "verified_on": "2026-10-10", "sources": [ "https://github.com/google-deepmind/graphcast", "https://developers.google.com/weathernext/guides/evals" ], "unverified": [ "release_date" ] }, { "id": "google-deepmind-genie", "name": "Genie", "level": "model", "family_id": "google-deepmind-genie-family", "developer": [ "google-deepmind" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-03", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First Genie (2D), Mar 2024.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Genie_(AI_model)" ], "unverified": [] }, { "id": "google-deepmind-genie-2", "name": "Genie 2", "level": "model", "family_id": "google-deepmind-genie-family", "developer": [ "google-deepmind" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-12", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Genie 2, 360p, 10 to 20 s worlds.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Genie_(AI_model)" ], "unverified": [] }, { "id": "google-deepmind-genie-3", "name": "Genie 3", "level": "model", "family_id": "google-deepmind-genie-family", "developer": [ "google-deepmind" ], "model_class": "world model", "modalities_in": [ "text", "action" ], "modalities_out": [ "video" ], "architecture": "autoregressive", "architecture_detail": "Generates frames from interactive input and previous frames; 720p, 24 fps, about 1 min visual memory", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-08-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "research only", "consumer app" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": "Google DeepMind real-time interactive world model, used to generate worlds for SIMA agents; consumer access via Project Genie.", "verified_on": "2026-10-10", "sources": [ "https://deepmind.google/discover/blog/genie-3-a-new-frontier-for-world-models/", "https://en.wikipedia.org/wiki/Genie_(AI_model)", "https://techcrunch.com/2025/08/05/deepmind-reveals-genie-3-a-world-model-that-could-be-the-key-to-reaching-agi", "https://deepmind.google/models/genie/" ], "unverified": [ "architecture" ] }, { "id": "google-deepmind-genie-family", "name": "Genie", "level": "family", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google DeepMind interactive world model line.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Genie_(AI_model)", "https://deepmind.google/discover/blog/genie-3-a-new-frontier-for-world-models/" ], "unverified": [] }, { "id": "google-deepmind-graphcast", "name": "GraphCast (WeatherNext 1 Graph)", "level": "model", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "GNN", "architecture_detail": "Graph neural network on multi-mesh; 0.25 deg, 37 levels; small 1 deg; operational 13 levels", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-sa-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Message-passing GNN weather model; beat HRES on 90% of tested targets; code Apache-2.0.", "verified_on": "2026-10-10", "sources": [ "https://github.com/google-deepmind/graphcast", "https://developers.google.com/weathernext/guides/evals", "https://sites.research.google/gr/weatherbench/" ], "unverified": [ "release_date" ] }, { "id": "google-deepmind-imagen", "name": "Imagen", "level": "family", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "Google text-to-image series, now described as discontinued; Imagen 4 API IDs shut down 17 Aug 2026.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Imagen_(text-to-image_model)", "https://kingy.ai/ai-launch-tracker/google-will-shut-down-three-imagen-4-api-models-august-17/", "https://deepmind.google/models/imagen/", "https://en.wikipedia.org/wiki/Adobe_Firefly" ], "unverified": [] }, { "id": "google-deepmind-imagen-4", "name": "Imagen 4", "level": "model", "family_id": "google-deepmind-imagen", "developer": [ "google-deepmind" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "shut down", "summary": "Last Imagen line; Gemini API IDs imagen-4.0-generate-001, ultra and fast shut down 17 Aug 2026, replaced by Nano Banana 2.1. Vertex AI status not confirmed.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Imagen_(text-to-image_model)", "https://kingy.ai/ai-launch-tracker/google-will-shut-down-three-imagen-4-api-models-august-17/", "https://deepmind.google/models/imagen/" ], "unverified": [] }, { "id": "google-deepmind-lyria", "name": "Lyria", "level": "model", "family_id": "google-deepmind-lyria-family", "developer": [ "google-deepmind" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First Lyria music model.", "verified_on": "2026-10-10", "sources": [ "https://deepmind.google/models/lyria/" ], "unverified": [] }, { "id": "google-deepmind-lyria-2", "name": "Lyria 2", "level": "model", "family_id": "google-deepmind-lyria-family", "developer": [ "google-deepmind" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Lyria 2 music model.", "verified_on": "2026-10-10", "sources": [ "https://deepmind.google/models/lyria/", "https://alphasignal.ai/news/artificial-analysis-music-arena-v1-1-crowns-suno-v6-across-1-000-harder-prompts" ], "unverified": [] }, { "id": "google-deepmind-lyria-3", "name": "Lyria 3", "level": "model", "family_id": "google-deepmind-lyria-family", "developer": [ "google-deepmind" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Lyria 3 music model.", "verified_on": "2026-10-10", "sources": [ "https://deepmind.google/models/lyria/" ], "unverified": [] }, { "id": "google-deepmind-lyria-3-5", "name": "Lyria 3.5", "level": "model", "family_id": "google-deepmind-lyria-family", "developer": [ "google-deepmind" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Generates vocals in different languages; Indonesian not stated", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google media model, noted in passing in the research notes.", "verified_on": "2026-10-10", "sources": [ "https://ai.google.dev/gemini-api/docs/models", "https://deepmind.google/models/lyria/", "https://alphasignal.ai/news/artificial-analysis-music-arena-v1-1-crowns-suno-v6-across-1-000-harder-prompts" ], "unverified": [] }, { "id": "google-deepmind-lyria-3-pro", "name": "Lyria 3 Pro", "level": "variant", "family_id": "google-deepmind-lyria-3", "developer": [ "google-deepmind" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Pro version of Lyria 3 on the AA Music Arena.", "verified_on": "2026-10-10", "sources": [ "https://alphasignal.ai/news/artificial-analysis-music-arena-v1-1-crowns-suno-v6-across-1-000-harder-prompts" ], "unverified": [] }, { "id": "google-deepmind-lyria-family", "name": "Lyria", "level": "family", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "music and audio", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google DeepMind music generation line.", "verified_on": "2026-10-10", "sources": [ "https://deepmind.google/models/lyria/" ], "unverified": [] }, { "id": "google-deepmind-lyria-realtime", "name": "Lyria RealTime", "level": "model", "family_id": "google-deepmind-lyria-family", "developer": [ "google-deepmind" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time Lyria model.", "verified_on": "2026-10-10", "sources": [ "https://deepmind.google/models/lyria/" ], "unverified": [] }, { "id": "google-deepmind-magenta-realtime-2", "name": "Magenta RealTime 2", "level": "model", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-06", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open sibling of Lyria (magenta-realtime-2 on Hugging Face).", "verified_on": "2026-10-10", "sources": [ "https://deepmind.google/models/lyria/", "https://huggingface.co/models?pipeline_tag=text-to-audio&sort=trending" ], "unverified": [] }, { "id": "google-deepmind-nano-banana-2-1", "name": "Nano Banana 2.1", "level": "model", "family_id": "google-nano-banana", "developer": [ "google-deepmind", "google" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": "Gemini-native", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-10-06", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "Cloud only; Gemini API and Batch API, ID gemini-nano-banana-2.1", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google media model, noted in passing in the research notes.", "verified_on": "2026-10-10", "sources": [ "https://ai.google.dev/gemini-api/docs/models", "https://www.orcarouter.ai/vi/blog/nano-banana-2-1-release-date", "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text", "https://arena.ai/leaderboard/text-to-image", "https://kingy.ai/ai-launch-tracker/google-will-shut-down-three-imagen-4-api-models-august-17/" ], "unverified": [] }, { "id": "google-deepmind-natural-forests-mtsvit", "name": "Natural Forests of the World 2020 model (MTSViT)", "level": "model", "family_id": null, "developer": [ "google-deepmind", "google", "wri", "iiasa" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": "Multi-modal temporal-spatial vision transformer on Sentinel-2 seasonal imagery plus elevation, slope and coordinates", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-11-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "cloud marketplace" ], "deployment_tiers": [], "footprint_note": "Map data on Figshare and Earth Engine", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Model behind the 10 m Natural Forests of the World 2020 map (92.2% accuracy vs an independent global validation set) for EUDR deforestation baselines; 2026 maps with a Tree Crops class planned.", "verified_on": "2026-10-10", "sources": [ "https://research.google/blog/separating-natural-forests-from-other-tree-cover-with-ai-for-deforestation-free-supply-chains/" ], "unverified": [] }, { "id": "google-deepmind-project-genie", "name": "Project Genie", "level": "model", "family_id": null, "developer": [ "google-deepmind", "google" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01-29", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Genie 3 web app for Google AI Ultra subscribers (18+), 60 s exploration, with Street View grounding added later.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Genie_(AI_model)", "https://techcrunch.com/2026/01/29/i-built-marshmallow-castles-in-googles-new-ai-world-generator-project-genie", "https://aiweekly.co/newsletters/ai-world-models/google-grounds-its-genie-world-model-in-20-years-of-street" ], "unverified": [] }, { "id": "google-deepmind-rt-2", "name": "RT-2", "level": "family", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "robot policy", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google DeepMind robotics transformer VLA.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2406.09246" ], "unverified": [] }, { "id": "google-deepmind-rt-2-x", "name": "RT-2-X", "level": "model", "family_id": "google-deepmind-rt-2", "developer": [ "google-deepmind" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": null, "architecture_detail": "VLM co-fine-tuned to output action tokens", "params_total": 55000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023", "release_date_basis": "notes", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "research only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google DeepMind 55B VLA; OpenVLA beats it by 16.5 points over 29 tasks.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2406.09246" ], "unverified": [ "release_date" ] }, { "id": "google-deepmind-veo", "name": "Veo", "level": "family", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "video generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google DeepMind video generation line.", "verified_on": "2026-10-10", "sources": [ "https://deepmind.google/models/veo/" ], "unverified": [] }, { "id": "google-deepmind-veo-3", "name": "Veo 3", "level": "model", "family_id": "google-deepmind-veo", "developer": [ "google-deepmind" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Previous Veo model; leads VBench-2.0.", "verified_on": "2026-10-10", "sources": [ "https://deepmind.google/models/veo/", "https://huggingface.co/spaces/Vchitect/VBench_Leaderboard" ], "unverified": [] }, { "id": "google-deepmind-veo-3-1", "name": "Veo 3.1", "level": "model", "family_id": "google-deepmind-veo", "developer": [ "google-deepmind" ], "model_class": "video generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "video", "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google media model, noted in passing in the research notes.", "verified_on": "2026-10-10", "sources": [ "https://ai.google.dev/gemini-api/docs/models", "https://deepmind.google/models/veo/", "https://artificialanalysis.ai/video/leaderboard/text-to-video", "https://arena.ai/leaderboard/text-to-video" ], "unverified": [] }, { "id": "google-deepmind-veo-3-1-fast", "name": "Veo 3.1 Fast", "level": "variant", "family_id": "google-deepmind-veo-3-1", "developer": [ "google-deepmind" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fast variant of Veo 3.1.", "verified_on": "2026-10-10", "sources": [ "https://deepmind.google/models/veo/" ], "unverified": [] }, { "id": "google-deepmind-veo-3-1-lite", "name": "Veo 3.1 Lite", "level": "variant", "family_id": "google-deepmind-veo-3-1", "developer": [ "google-deepmind" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-03", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Lite variant of Veo 3.1.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/video/leaderboard/text-to-video" ], "unverified": [] }, { "id": "google-deepmind-weathernext-1-gen", "name": "WeatherNext 1 Gen (GenCast)", "level": "model", "family_id": null, "developer": [ "google-deepmind", "google" ], "model_class": "weather", "modalities_in": [], "modalities_out": [], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-12", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "GenCast diffusion weather model, predecessor in the WeatherNext line.", "verified_on": "2026-10-10", "sources": [ "https://developers.google.com/weathernext/guides/models" ], "unverified": [] }, { "id": "google-deepmind-weathernext-1-graph", "name": "WeatherNext 1 Graph (GraphCast)", "level": "model", "family_id": null, "developer": [ "google-deepmind", "google" ], "model_class": "weather", "modalities_in": [], "modalities_out": [], "architecture": "GNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-11", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "GraphCast GNN weather model, predecessor in the WeatherNext line.", "verified_on": "2026-10-10", "sources": [ "https://developers.google.com/weathernext/guides/models" ], "unverified": [] }, { "id": "google-deepmind-weathernext-2", "name": "WeatherNext 2 (FGN)", "level": "model", "family_id": null, "developer": [ "google-deepmind", "google" ], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "other", "architecture_detail": "Functional Generative Network (noise injected into architecture), ensemble; hourly output", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Google docs give June 2025 as the model-version date", "context_length": null, "release_date": "2025-11-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "cloud marketplace", "gated preview" ], "deployment_tiers": [ "hyperscale cluster" ], "footprint_note": "Hundreds of members in under 1 min on one TPU", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "Google weather ensemble model; beats WeatherNext Gen on 99.9% of variables and lead times (CRPS). Google docs give June 2025 as model-version date; WeatherNext 3 (Aug 2026) is now recommended.", "verified_on": "2026-10-10", "sources": [ "https://blog.google/technology/google-deepmind/weathernext-2/", "https://developers.google.com/weathernext/guides/evals", "https://developers.google.com/weathernext/guides/models", "https://developers.google.com/weathernext/guides/models-wn2", "https://developers.google.com/weathernext/guides/deprecation", "https://blog.google/technology/google-deepmind/weathernext-2/" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "google-deepmind-weathernext-3", "name": "WeatherNext 3", "level": "model", "family_id": null, "developer": [ "google-deepmind", "google" ], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "dense transformer", "architecture_detail": "FGN mesh transformer, 64-member ensemble; 0.05 deg stations, 0.1 deg surface, 0.25 deg pressure levels, hourly", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "gdm-experimental-data-terms", "access_routes": [ "cloud marketplace" ], "deployment_tiers": [], "footprint_note": "BigQuery, Earth Engine, Cloud Storage. Real-time data under GDM Experimental Data Terms; historical (over 1 hour old) CC BY 4.0.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google flagship operational weather model, recommended for all new projects.", "verified_on": "2026-10-10", "sources": [ "https://developers.google.com/weathernext/guides/models", "https://developers.google.com/weathernext/guides/deprecation", "https://developers.google.com/weathernext/guides/models-wn2", "https://blog.google/technology/google-deepmind/weathernext-2/" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "google-deepvariant", "name": "DeepVariant", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "scientific", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Variant calling model attributed to Google; unverified (page not read).", "verified_on": "2026-10-10", "sources": [ "https://github.com/ArcInstitute/evo2" ], "unverified": [ "developer", "summary" ] }, { "id": "google-embeddinggemma-2", "name": "EmbeddingGemma 2", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "embedding", "modalities_in": [ "text", "image", "audio" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Gemma 4 architecture", "params_total": 740000000, "params_active": null, "params_checkpoint": null, "params_note": "740M total (270M text path; vision and audio encoders)", "context_length": 8192, "release_date": "2026-10-07", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "100+ languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google's embedding model on the Gemma 4 architecture, 768-dimension Matryoshka (512, 256, 128).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/google/embeddinggemma-2" ], "unverified": [] }, { "id": "google-embeddinggemma-300m", "name": "EmbeddingGemma 300M", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "300M", "context_length": null, "release_date": "2025-07-17", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "CPU server", "laptop", "mobile or NPU" ], "footprint_note": "Ollama GGUF build", "runtimes": [ "ollama" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small model listed in the Ollama library for CPU, laptop and phone use.", "verified_on": "2026-10-10", "sources": [ "https://ollama.com/library", "https://huggingface.co/google/embeddinggemma-300m" ], "unverified": [] }, { "id": "google-functiongemma-270m", "name": "FunctionGemma 270M", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "270M", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "CPU server", "laptop", "mobile or NPU" ], "footprint_note": "Ollama GGUF build", "runtimes": [ "ollama" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small model listed in the Ollama library for CPU, laptop and phone use.", "verified_on": "2026-10-10", "sources": [ "https://ollama.com/library" ], "unverified": [] }, { "id": "google-gemini-3-1-flash-tts", "name": "Gemini 3.1 Flash TTS", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier Gemini TTS model.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/text-to-speech/leaderboard" ], "unverified": [] }, { "id": "google-gemini-3-8-flash-lite-tts", "name": "Gemini 3.8 Flash-Lite TTS", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-09", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "100+ languages", "indonesian_support": "listed by developer", "regional_languages": [ "Javanese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Google TTS model gemini-3.8-flash-lite-tts.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/text-to-speech/leaderboard", "https://ai.google.dev/gemini-api/docs/speech-generation" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "google-gemini-3-8-flash-tts", "name": "Gemini 3.8 Flash TTS", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-09", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "130+ languages", "indonesian_support": "listed by developer", "regional_languages": [ "Javanese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Google TTS model gemini-3.8-flash-tts, 5th on the Artificial Analysis TTS arena.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/text-to-speech/leaderboard", "https://ai.google.dev/gemini-api/docs/speech-generation" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "google-gemini-3-flash", "name": "Gemini 3 Flash", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google proprietary LLM, used as a reference row in the OmniDocBench v1.6 table.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2606.03264" ], "unverified": [] }, { "id": "google-gemini-nano", "name": "Gemini Nano", "level": "model", "family_id": "google-deepmind-gemini", "developer": [ "google" ], "model_class": "edge LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "On-device Gemini model that Honeywell plans to use on edge devices.", "verified_on": "2026-10-10", "sources": [ "https://www.designnews.com/automation/honeywell-uses-google-cloud-ai-to-accelerate-autonomous-operations" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "google-gemini-nano-banana-2-1", "name": "Gemini Nano Banana 2.1", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google image model shown at #6 in Arena Text-to-Image.", "verified_on": "2026-10-10", "sources": [ "https://arena.ai/leaderboard" ], "unverified": [] }, { "id": "google-gemini-omni", "name": "Gemini Omni", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "video generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google unified video generation pipeline combining Gemini, Nano Banana and Veo.", "verified_on": "2026-10-10", "sources": [ "https://framia.converge.ai/page/en-US/news/gemini-omni-ai-video-generation-model" ], "unverified": [] }, { "id": "google-gemini-omni-flash-1-1", "name": "Gemini Omni Flash 1.1", "level": "model", "family_id": "google-gemini-omni", "developer": [ "google" ], "model_class": "video generation", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "video", "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-08", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Update of Gemini Omni Flash, leading LMArena Text-to-Video but 8th on Artificial Analysis.", "verified_on": "2026-10-10", "sources": [ "https://arena.ai/leaderboard/text-to-video", "https://artificialanalysis.ai/video/leaderboard/text-to-video", "https://framia.converge.ai/page/en-US/news/gemini-omni-ai-video-generation-model" ], "unverified": [] }, { "id": "google-gemma", "name": "Gemma", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Google open model family; base of Gemma-SEA-LION-v4.5-E2B-IT (exact base not stated).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/aisingapore" ], "unverified": [] }, { "id": "google-gemma-2", "name": "Gemma 2", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": "gemma-terms", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/google/models?search=shieldgemma" ], "unverified": [] }, { "id": "google-gemma-2-2b", "name": "Gemma 2 2B", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": 2000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-07-16", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Open model used for Gemma Scope SAEs on all layers and for circuit-tracer demos.", "verified_on": "2026-10-10", "sources": [ "https://anthropic.com/research/open-source-circuit-tracing", "https://arxiv.org/abs/2408.05147", "https://huggingface.co/google/gemma-2-2b" ], "unverified": [] }, { "id": "google-gemma-2-2b-it", "name": "Gemma2-2b-IT", "level": "model", "family_id": "google-gemma-2", "developer": [ "google" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Text decoder used in MERaLiON-2-3B.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/MERaLiON/MERaLiON-2-10B" ], "unverified": [] }, { "id": "google-gemma-2-9b", "name": "Gemma 2 9B", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 9000000000, "params_active": null, "params_checkpoint": null, "params_note": "9B (inferred from gemma2-9b-cpt name)", "context_length": null, "release_date": "2024-06-24", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google Gemma 2 model inferred as the root of the SEA-LION v3 and Sahabat-AI chain.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/models?search=sahabat", "https://huggingface.co/api/models/GoToCompany/gemma2-9b-cpt-sahabatai-v1-base", "https://huggingface.co/GoToCompany", "https://arxiv.org/abs/2408.05147", "https://huggingface.co/google/gemma-2-9b" ], "unverified": [ "params_note" ] }, { "id": "google-gemma-2-9b-it", "name": "Gemma2-9b-IT", "level": "model", "family_id": "google-gemma-2", "developer": [ "google" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Text decoder used in MERaLiON-2-10B.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/MERaLiON/MERaLiON-2-10B" ], "unverified": [] }, { "id": "google-gemma-3", "name": "Gemma 3", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": "gemma-terms", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google Gemma 3 family; small sizes in Ollama and widely used bases on Hugging Face.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/google/models?search=shieldgemma", "https://ollama.com/library", "https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct", "https://arxiv.org/html/2507.05201v2", "https://huggingface.co/google/medgemma-27b-text-it", "https://blog.google/technology/developers/gemma-3/", "https://huggingface.co/google/gemma-3-27b-it" ], "unverified": [] }, { "id": "google-gemma-3-1b", "name": "Gemma 3 1B", "level": "model", "family_id": "google-gemma-3", "developer": [ "google" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "1B", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "CPU server", "laptop", "mobile or NPU" ], "footprint_note": "Ollama GGUF build", "runtimes": [ "ollama" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small model listed in the Ollama library for CPU, laptop and phone use.", "verified_on": "2026-10-10", "sources": [ "https://ollama.com/library" ], "unverified": [] }, { "id": "google-gemma-3-270m", "name": "Gemma 3 270M", "level": "model", "family_id": "google-gemma-3", "developer": [ "google" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "270M", "context_length": null, "release_date": "2025-08-05", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "CPU server", "laptop", "mobile or NPU" ], "footprint_note": "Ollama GGUF build", "runtimes": [ "ollama" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small model listed in the Ollama library for CPU, laptop and phone use.", "verified_on": "2026-10-10", "sources": [ "https://ollama.com/library", "https://huggingface.co/google/gemma-3-270m" ], "unverified": [] }, { "id": "google-gemma-3-27b", "name": "Gemma 3 27B", "level": "model", "family_id": "google-gemma-3", "developer": [ "google" ], "model_class": "multimodal LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": 27000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Base model of MedGemma 27B.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2507.05201v2" ], "unverified": [] }, { "id": "google-gemma-3-27b-it", "name": "gemma-3-27b-it", "level": "model", "family_id": "google-gemma-3", "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "HF model tree 2026-10-10: 452 finetunes, 278 adapters, 145 quantizations, 5 merges", "context_length": null, "release_date": "2025-03-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Gemma 3 instruct model; widely used base on Hugging Face.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct", "https://huggingface.co/api/models/aisingapore/Gemma-SEA-LION-v4-27B", "https://blog.google/technology/developers/gemma-3/" ], "unverified": [] }, { "id": "google-gemma-3-4b", "name": "Gemma 3 4B", "level": "model", "family_id": "google-gemma-3", "developer": [ "google" ], "model_class": "multimodal LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": 4000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Base model of MedGemma 4B; scores 50.7 on MedQA in the MedGemma report.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2507.05201v2", "https://blog.google/technology/developers/gemma-3/", "https://arxiv.org/abs/2503.19786" ], "unverified": [] }, { "id": "google-gemma-3-4b-it", "name": "Gemma-3-4B-it", "level": "model", "family_id": "google-gemma-3", "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 4000000000, "params_active": null, "params_checkpoint": null, "params_note": "HF model tree 2026-10-10: 807 finetunes, 498 adapters, 490 quantizations, 3 merges", "context_length": null, "release_date": "2025-03-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": "gemma-terms", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Gemma 3 instruct model; widely used base on Hugging Face.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/nvidia/Nemotron-3.5-Content-Safety", "https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct", "https://huggingface.co/api/models/aisingapore/Gemma-SEA-LION-v4-27B", "https://blog.google/technology/developers/gemma-3/" ], "unverified": [] }, { "id": "google-gemma-3n", "name": "Gemma 3n", "level": "family", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "edge LLM", "modalities_in": [ "text", "image", "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "MatFormer architecture, Per-Layer Embedding (PLE) caching, conditional loading of vision and audio parameters, MobileNet-V5 vision encoder", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 32000, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU", "laptop", "offline capable" ], "footprint_note": null, "runtimes": [], "languages_note": "Trained on over 140 languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google's on-device multimodal open model family for phones, laptops and tablets; predecessor to the Gemma 4 E-series.", "verified_on": "2026-10-10", "sources": [ "https://deepmind.google/models/gemma/gemma-3n/", "https://ai.google.dev/gemma/docs/gemma-3n", "https://ai.google.dev/gemma/docs/gemma-3n/model_card" ], "unverified": [] }, { "id": "google-gemma-3n-e2b", "name": "Gemma 3n E2B", "level": "model", "family_id": "google-gemma-3n", "developer": [ "google-deepmind" ], "model_class": "edge LLM", "modalities_in": [ "text", "image", "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "MatFormer; MobileNet-V5 vision encoder; audio input at 6.25 tokens/s", "params_total": null, "params_active": null, "params_checkpoint": 5439438272, "params_note": "E2B-it raw parameter total on HF. With parameter skipping and PLE caching it runs at an effective memory load of just under 2 billion (1.91B) parameters; standard execution loads over 5 billion.", "context_length": 32000, "release_date": "2025-06-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU", "laptop", "offline capable" ], "footprint_note": "Footprint in GB and measured tok/s on a named device unverified; official LiteRT-LM build google/gemma-3n-E2B-it-litert-lm", "runtimes": [ "litert" ], "languages_note": "Trained on over 140 languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Gemma 3n E2B on-device multimodal model for phones, laptops and tablets.", "verified_on": "2026-10-10", "sources": [ "https://ai.google.dev/gemma/docs/gemma-3n", "https://ai.google.dev/gemma/docs/gemma-3n/model_card", "https://huggingface.co/api/models/google/gemma-3n-E2B-it", "https://huggingface.co/api/models?author=google&search=gemma-3n", "https://deepmind.google/models/gemma/gemma-3n/", "https://huggingface.co/google/gemma-3n-E2B" ], "unverified": [ "footprint_note" ] }, { "id": "google-gemma-3n-e4b", "name": "Gemma 3n E4B", "level": "model", "family_id": "google-gemma-3n", "developer": [ "google-deepmind" ], "model_class": "edge LLM", "modalities_in": [ "text", "image", "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "MatFormer; MobileNet-V5 vision encoder; audio input at 6.25 tokens/s", "params_total": null, "params_active": null, "params_checkpoint": 7849978192, "params_note": "E4B-it raw parameter total on HF. Published on HF 3 Jun 2025 (E4B-it).", "context_length": 32000, "release_date": "2025-06-03", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU", "laptop", "offline capable" ], "footprint_note": "Footprint in GB and measured tok/s on a named device unverified", "runtimes": [], "languages_note": "Trained on over 140 languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Gemma 3n E4B on-device multimodal model for phones, laptops and tablets.", "verified_on": "2026-10-10", "sources": [ "https://ai.google.dev/gemma/docs/gemma-3n", "https://ai.google.dev/gemma/docs/gemma-3n/model_card", "https://huggingface.co/api/models/google/gemma-3n-E4B-it", "https://deepmind.google/models/gemma/gemma-3n/" ], "unverified": [ "footprint_note" ] }, { "id": "google-gemma-4", "name": "Gemma 4", "level": "family", "family_id": null, "developer": [ "google", "google-deepmind" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Released in E2B, E4B, 31B and 26B A4B sizes (Google release log); release log date 31 Mar 2026 vs announcement 2 Apr 2026 (open question)", "context_length": null, "release_date": "2026-04-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google's open-weight model family, related to Gemini.", "verified_on": "2026-10-10", "sources": [ "https://blog.google/innovation-and-ai/technology/developers-tools/gemma-4/", "https://ai.google.dev/gemma/docs/releases", "https://www.orcarouter.ai/es/blog/gemini-4-argon", "https://huggingface.co/google/gemma-4-31B-it", "https://huggingface.co/google", "https://huggingface.co/api/models/google/gemma-4-31B-it/commits/main", "https://datanorth.ai/news/google-releases-gemma-4-open-models", "https://www.mindstudio.ai/blog/gemma-4-e2b-e4b-edge-models-phone-local" ], "unverified": [] }, { "id": "google-gemma-4-12b", "name": "Gemma 4 12B", "level": "model", "family_id": "google-gemma-4", "developer": [ "google" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "audio" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "\"Unified\", encoder-free", "params_total": 12000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 256000, "release_date": "2026-05-23", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google's Gemma 4 12B unified, encoder-free model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/google/gemma-4-31B-it", "https://huggingface.co/google", "https://huggingface.co/google/gemma-4-12B" ], "unverified": [] }, { "id": "google-gemma-4-26b-a4b", "name": "Gemma 4 26B-A4B", "level": "model", "family_id": "google-gemma-4", "developer": [ "google" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "MoE (128 experts, 8 active + 1 shared)", "params_total": 25200000000, "params_active": 3800000000, "params_checkpoint": null, "params_note": null, "context_length": 256000, "release_date": "2026-04-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": "140+ pretraining languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google's Gemma 4 MoE text and image model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/google/gemma-4-31B-it", "https://huggingface.co/google", "https://ai.google.dev/gemma/docs/releases" ], "unverified": [] }, { "id": "google-gemma-4-31b", "name": "Gemma 4 31B", "level": "model", "family_id": "google-gemma-4", "developer": [ "google", "google-deepmind" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "Dense; hybrid local sliding-window + global attention, p-RoPE", "params_total": 30700000000, "params_active": null, "params_checkpoint": null, "params_note": "31B (flagship of the family)", "context_length": 256000, "release_date": "2026-04-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU", "consumer GPU" ], "footprint_note": "Single datacentre GPU; 4-bit on consumer GPU", "runtimes": [], "languages_note": null, "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google's 31B dense text and image model, top open model on SEA-HELM (18 Sep 2026); tech report 2 Jul 2026. 31B-it: finetunes 287, adapters 331, merges 38, quants 325.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/google/gemma-4-31B-it", "https://huggingface.co/google", "https://leaderboard.sea-lion.ai/detailed/SEA", "https://leaderboard.sea-lion.ai/detailed/ID", "https://blog.google/innovation-and-ai/technology/developers-tools/gemma-4/", "https://huggingface.co/api/models/google/gemma-4-31B-it/commits/main", "https://datanorth.ai/news/google-releases-gemma-4-open-models" ], "unverified": [], "indonesian_measured": [ { "benchmark": "SEA-HELM Indonesian leaderboard (ID aggregate, first score column of the Indonesian Competency and Task Scores table)", "score": "79.75 (95% CI -1.25 to +1.20)", "metric": "score 0 to 100, average of 2000 bootstraps with 95% CI", "run_by": "AI Singapore (SEA-HELM leaderboard maintainer)", "url": "https://leaderboard.sea-lion.ai/detailed/ID" } ] }, { "id": "google-gemma-4-e2b", "name": "Gemma 4 E2B", "level": "model", "family_id": "google-gemma-4", "developer": [ "google", "google-deepmind" ], "model_class": "edge LLM", "modalities_in": [ "text", "image", "audio" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "Dense with Per-Layer Embeddings; audio capped at 30 s", "params_total": 5100000000, "params_active": 2300000000, "params_checkpoint": null, "params_note": "2.3B effective (5.1B with embeddings)", "context_length": 128000, "release_date": "2026-04-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "laptop", "mobile or NPU", "offline capable" ], "footprint_note": "Under 1.5 GB at 2-bit (Raspberry Pi 5); about 1.3 GB at Q4_K_M. Raspberry Pi 5: 133 tok/s prefill, 7.6 tok/s decode. Snapdragon 8 Gen 3: 20 to 35 tok/s (third-party estimate). Android via AICore Developer Preview. LiteRT-LM packages exist", "runtimes": [ "litert" ], "languages_note": null, "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google's small Gemma 4 model for CPU and edge use. E2B-it has 337 derivatives, including Gemma-SEA-LION-v4.5-E2B.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/google/gemma-4-31B-it", "https://huggingface.co/google", "https://huggingface.co/aisingapore/Gemma-SEA-LION-v4.5-E2B-IT", "https://leaderboard.sea-lion.ai/detailed/SEA", "https://leaderboard.sea-lion.ai/detailed/ID", "https://ai.google.dev/gemma/docs/releases", "https://datanorth.ai/news/google-releases-gemma-4-open-models", "https://www.mindstudio.ai/blog/gemma-4-e2b-e4b-edge-models-phone-local", "https://huggingface.co/huggingworld/gemma-4-E2B-it-litert-lm", "https://blog.google/innovation-and-ai/technology/developers-tools/gemma-4/" ], "unverified": [ "architecture_detail" ], "indonesian_measured": [ { "benchmark": "SEA-HELM Indonesian leaderboard (ID aggregate, first score column of the Indonesian Competency and Task Scores table)", "score": "64.13 (95% CI -1.39 to +1.41)", "metric": "score 0 to 100, average of 2000 bootstraps with 95% CI", "run_by": "AI Singapore (SEA-HELM leaderboard maintainer)", "url": "https://leaderboard.sea-lion.ai/detailed/ID" } ] }, { "id": "google-gemma-4-e4b", "name": "Gemma 4 E4B", "level": "model", "family_id": "google-gemma-4", "developer": [ "google", "google-deepmind" ], "model_class": "edge LLM", "modalities_in": [ "text", "image", "audio" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "Dense with Per-Layer Embeddings", "params_total": 8000000000, "params_active": 4500000000, "params_checkpoint": null, "params_note": "4.5B effective (8B with embeddings)", "context_length": 128000, "release_date": "2026-04-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "laptop", "mobile or NPU", "offline capable" ], "footprint_note": "About 2.5 GB at Q4_K_M; 2 to 3 GB at 4-bit. Phones with 6 GB+ RAM, iPhone 14 Pro+, Apple Silicon. Snapdragon 8 Gen 3: 12 to 20 tok/s (third-party estimate)", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google's small Gemma 4 model for CPU and edge use.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/google/gemma-4-31B-it", "https://huggingface.co/google", "https://ai.google.dev/gemma/docs/releases", "https://datanorth.ai/news/google-releases-gemma-4-open-models", "https://www.mindstudio.ai/blog/gemma-4-e2b-e4b-edge-models-phone-local", "https://blog.google/innovation-and-ai/technology/developers-tools/gemma-4/" ], "unverified": [] }, { "id": "google-med-palm-2", "name": "Med-PaLM 2", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023", "release_date_basis": "notes", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Predecessor line to AMIE, 86.5% on MedQA (2023); unverified in the notes.", "verified_on": "2026-10-10", "sources": [ "https://blog.google/innovation-and-ai/models-and-research/google-research/amie-video-consultations/" ], "unverified": [ "summary", "release_date" ] }, { "id": "google-medgemma", "name": "MedGemma", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "multimodal LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights restricted", "licence": "hai-def-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Google's medical model family built on Gemma 3, with 4B multimodal, 27B text-only and 27B multimodal variants plus MedGemma 1.5 4B. Outputs are not for direct clinical decision-making per the model card.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2604.05081", "https://arxiv.org/html/2507.05201v2", "https://huggingface.co/google/medgemma-27b-text-it" ], "unverified": [] }, { "id": "google-medgemma-1-5-4b", "name": "MedGemma 1.5 4B", "level": "model", "family_id": "google-medgemma", "developer": [ "google" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "3D" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 4000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04-06", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "hai-def-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": "On premises or in the cloud (open weights).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Update of MedGemma 4B (arXiv 2604.05081, 6 Apr 2026) adding 3D CT and MRI, whole-slide pathology, anatomical localization, longitudinal chest X-ray and document understanding. Reported gains over MedGemma 1 4B: MedQA +5%, EHRQA +22%, WSI macro F1 +47%, 3D MRI +11% (absolute or relative not stated).", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2604.05081" ], "unverified": [] }, { "id": "google-medgemma-27b-multimodal", "name": "MedGemma 27B multimodal", "level": "model", "family_id": "google-medgemma", "developer": [ "google" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 27000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights restricted", "licence": "hai-def-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": "On premises or in the cloud (open weights).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "27B multimodal MedGemma variant; its evaluation was still ongoing in the July 2025 technical report.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2507.05201v2" ], "unverified": [] }, { "id": "google-medgemma-27b-text", "name": "MedGemma 27B text", "level": "model", "family_id": "google-medgemma", "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 27000000000, "params_active": null, "params_checkpoint": null, "params_note": "Hugging Face model card created 20 May 2025.", "context_length": null, "release_date": "2025-05-19", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "hai-def-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": "On premises or in the cloud (open weights).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "27B text-only medical LLM built on Gemma 3 27B. Scores 87.7 on MedQA (0-shot with test-time scaling) per the MedGemma Technical Report.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2507.05201v2", "https://huggingface.co/google/medgemma-27b-text-it" ], "unverified": [] }, { "id": "google-medgemma-4b", "name": "MedGemma 4B", "level": "model", "family_id": "google-medgemma", "developer": [ "google" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 4000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights restricted", "licence": "hai-def-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": "On premises or in the cloud (open weights).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "4B multimodal medical model built on Gemma 3 4B, used for radiology report generation and medical QA. Dermatology images among its inputs is unverified in the notes.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2507.05201v2", "https://huggingface.co/google/medgemma-27b-text-it" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "google-medsiglip", "name": "MedSigLIP", "level": "model", "family_id": "google-medgemma", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": 400000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "400M medical image encoder derived from SigLIP-400M, part of the MedGemma release.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2507.05201v2", "https://huggingface.co/google/medgemma-27b-text-it" ], "unverified": [] }, { "id": "google-model-armor", "name": "Google Model Armor and Vertex safety filters", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "guard", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google safety filters (docs fetch failed); known filter families: responsible-AI categories, prompt injection and jailbreak, malicious URLs, sensitive data.", "verified_on": "2026-10-10", "sources": [ "https://cloud.google.com/model-armor/docs/overview" ], "unverified": [ "summary", "languages_note" ] }, { "id": "google-mt5", "name": "mT5", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/indonlp" ], "unverified": [] }, { "id": "google-nano-banana", "name": "Nano Banana", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google's Gemini-native image generation line (Nano Banana 2, 2.1, Pro).", "verified_on": "2026-10-10", "sources": [ "https://www.orcarouter.ai/vi/blog/nano-banana-2-1-release-date", "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text", "https://arena.ai/leaderboard/text-to-image" ], "unverified": [] }, { "id": "google-nano-banana-2", "name": "Nano Banana 2 (Gemini 3.1 Flash Image)", "level": "model", "family_id": "google-nano-banana", "developer": [ "google" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-02", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "Google image model (Gemini 3.1 Flash Image), Feb 2026; shuts down 29 Oct 2026.", "verified_on": "2026-10-10", "sources": [ "https://www.orcarouter.ai/vi/blog/nano-banana-2-1-release-date", "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text" ], "unverified": [] }, { "id": "google-nano-banana-2-lite", "name": "Nano Banana 2 Lite", "level": "model", "family_id": "google-nano-banana", "developer": [ "google" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-06", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google image model listed on the Artificial Analysis Text-to-Image Arena.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text" ], "unverified": [] }, { "id": "google-nano-banana-pro", "name": "Nano Banana Pro (Gemini 3 Pro Image)", "level": "model", "family_id": "google-nano-banana", "developer": [ "google" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-11", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google image model (Gemini 3 Pro Image), Nov 2025, still top tier. C2PA metadata on its images from the Gemini app, Vertex AI and Google Ads since the week of 20 Nov 2025.", "verified_on": "2026-10-10", "sources": [ "https://www.orcarouter.ai/vi/blog/nano-banana-2-1-release-date", "https://arena.ai/leaderboard/text-to-image", "https://blog.google/technology/ai/ai-image-verification-gemini-app/" ], "unverified": [] }, { "id": "google-palm-2", "name": "PaLM 2", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Google LLM behind the June 2024 Google Translate 110-language expansion.", "verified_on": "2026-10-10", "sources": [ "https://blog.google/products/translate/google-translate-new-languages-2024/" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "google-shieldgemma-2-4b", "name": "ShieldGemma 2 4B", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "guard", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 4000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-04", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English (unverified)", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google's image safety guard (sexual, dangerous, violence) on Gemma 3; no Gemma 4 based ShieldGemma found.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/google/models?search=shieldgemma" ], "unverified": [ "languages_note" ] }, { "id": "google-shieldgemma-27b", "name": "ShieldGemma 27B", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "guard", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 27000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English (unverified)", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google's text-policy guard model on Gemma 2.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/google/models?search=shieldgemma" ], "unverified": [ "languages_note" ] }, { "id": "google-shieldgemma-2b", "name": "ShieldGemma 2B", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "guard", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 2000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English (unverified)", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google's text-policy guard model on Gemma 2.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/google/models?search=shieldgemma" ], "unverified": [ "languages_note" ] }, { "id": "google-shieldgemma-9b", "name": "ShieldGemma 9B", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "guard", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 9000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English (unverified)", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google's text-policy guard model on Gemma 2.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/google/models?search=shieldgemma" ], "unverified": [ "languages_note" ] }, { "id": "google-stride-w-synapse", "name": "STRIDE_w_Synapse", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google Cloud AI Research agentic forecasting system; rank 1 on the GIFT-Eval leaderboard rebuilt from its raw results.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/spaces/Salesforce/GIFT-Eval" ], "unverified": [] }, { "id": "google-timesfm", "name": "TimesFM", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google Research time-series foundation model family (1.0, 2.0, 2.5, 3.0).", "verified_on": "2026-10-10", "sources": [ "https://thenewstack.io/google-timesfm-3-multivariate-forecasting/", "https://www.beri.net/article/timesfm-vs-chronos-vs-moirai-time-series-foundation-model-licence-bigquery-sagemaker-cost-2026" ], "unverified": [] }, { "id": "google-timesfm-1-0", "name": "TimesFM 1.0", "level": "model", "family_id": "google-timesfm", "developer": [ "google" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First TimesFM version.", "verified_on": "2026-10-10", "sources": [ "https://thenewstack.io/google-timesfm-3-multivariate-forecasting/", "https://huggingface.co/google/timesfm-2.5-200m-pytorch" ], "unverified": [] }, { "id": "google-timesfm-2-0", "name": "TimesFM 2.0", "level": "model", "family_id": "google-timesfm", "developer": [ "google" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier TimesFM version.", "verified_on": "2026-10-10", "sources": [ "https://thenewstack.io/google-timesfm-3-multivariate-forecasting/", "https://huggingface.co/google/timesfm-2.5-200m-pytorch" ], "unverified": [] }, { "id": "google-timesfm-2-5", "name": "TimesFM 2.5", "level": "model", "family_id": "google-timesfm", "developer": [ "google" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "dense transformer", "architecture_detail": "Decoder-only", "params_total": 200000000, "params_active": null, "params_checkpoint": null, "params_note": "timesfm-2.5-200m-pytorch", "context_length": null, "release_date": "2025-09-02", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "single datacentre GPU", "CPU server" ], "footprint_note": "Cloud (BigQuery univariate forecasting path, unverified) or local. Hugging Face update 2 Oct 2025.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "TimesFM version behind BigQuery AI.FORECAST that 3.0 is planned to replace (implied).", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2605.20119", "https://thenewstack.io/google-timesfm-3-multivariate-forecasting/", "https://www.beri.net/article/timesfm-vs-chronos-vs-moirai-time-series-foundation-model-licence-bigquery-sagemaker-cost-2026", "https://huggingface.co/api/models/google/timesfm-2.5-200m-pytorch", "https://huggingface.co/google/timesfm-2.5-200m-pytorch", "https://github.com/google-research/timesfm" ], "unverified": [] }, { "id": "google-timesfm-3-0", "name": "TimesFM 3.0", "level": "model", "family_id": "google-timesfm", "developer": [ "google" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "dense transformer", "architecture_detail": "Decoder-only transformer, 32-point patches; alternating within-series causal attention and cross-series attention; masked full-horizon single-pass decoding; native multivariate + covariates", "params_total": 330000000, "params_active": null, "params_checkpoint": null, "params_note": "about 330M; trained on more than 1 trillion time points", "context_length": null, "release_date": "2026-08-31", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights restricted", "licence": "timesfm-non-commercial-license-v1-0", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": "Production use only via BigQuery; PyPI timesfm 3.0.2 on 9 Sep 2026", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google Research multivariate time-series foundation model; best single zero-shot model (rank 14 of 143) on the GIFT-Eval leaderboard rebuilt from its raw results. Code Apache, weights non-commercial.", "verified_on": "2026-10-10", "sources": [ "https://thenewstack.io/google-timesfm-3-multivariate-forecasting/", "https://www.beri.net/article/timesfm-vs-chronos-vs-moirai-time-series-foundation-model-licence-bigquery-sagemaker-cost-2026", "https://huggingface.co/spaces/Salesforce/GIFT-Eval", "https://pypi.org/pypi/timesfm/json", "https://datanorth.ai/nl/nieuws/google-lanceert-timesfm-3", "https://www.llmreference.com/model/timesfm-3" ], "unverified": [] }, { "id": "goto-gemma2-9b-cpt-sahabatai-v1", "name": "gemma2-9b-cpt-sahabatai-v1", "level": "model", "family_id": "goto-sahabat-ai-v1", "developer": [ "goto", "indosat", "ai-singapore" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 9000000000, "params_active": null, "params_checkpoint": null, "params_note": "9B to 10B", "context_length": null, "release_date": "2024-11-06", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": "Open weights; on-prem or Indonesian territory; sovereign cloud GPU Merdeka by Lintasarta", "runtimes": [], "languages_note": "Indonesian + regional languages", "indonesian_support": "listed by developer", "regional_languages": [ "Javanese", "Sundanese", "Batak Toba", "Balinese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Sahabat-AI v1 continued-pretraining model (base and instruct).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/GoToCompany", "https://huggingface.co/models?search=sahabat" ], "unverified": [ "licence", "indonesian_measured", "indonesia_hosting" ] }, { "id": "goto-gemma2-9b-cpt-sahabatai-v1-base", "name": "gemma2-9b-cpt-sahabatai-v1-base (Sahabat-AI v1 base)", "level": "model", "family_id": null, "developer": [ "goto" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "9B (per name)", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Indonesian Sahabat-AI v1 base model, continued pretraining on SEA-LION v3; GoTo with Indosat (partner unverified).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/api/models/GoToCompany/gemma2-9b-cpt-sahabatai-v1-base" ], "unverified": [ "developer" ] }, { "id": "goto-gemma2-9b-cpt-sahabatai-v1-instruct", "name": "gemma2-9b-cpt-sahabatai-v1-instruct (Sahabat-AI v1 instruct)", "level": "model", "family_id": null, "developer": [ "goto" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "9B (per name)", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Instruct version of the Indonesian Sahabat-AI v1 model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/api/models/GoToCompany/gemma2-9b-cpt-sahabatai-v1-base" ], "unverified": [] }, { "id": "goto-llama-sahabat-ai-v2-70b-it", "name": "Llama-Sahabat-AI-v2-70B-IT", "level": "model", "family_id": "goto-sahabat-ai", "developer": [ "goto", "indosat", "ai-singapore" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 71000000000, "params_active": null, "params_checkpoint": null, "params_note": "70B (per name); no base_model declared (relation finetune)", "context_length": null, "release_date": "2025-06-02", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "llama-3-1-community", "access_routes": [ "open weights download" ], "deployment_tiers": [ "multi-GPU node" ], "footprint_note": "About 140 GB in BF16", "runtimes": [], "languages_note": "English, Indonesian, Javanese, Sundanese, Batak Toba, Balinese", "indonesian_support": "listed by developer", "regional_languages": [ "Javanese", "Sundanese", "Batak Toba", "Balinese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Indonesian 70B instruct model by GoTo and Indosat, developed with AI Singapore; evaluated on IndoMMLU, SEA-HELM and SEA-IFEval (scores not listed on card).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/GoToCompany", "https://huggingface.co/GoToCompany/Llama-Sahabat-AI-v2-70B-IT", "https://huggingface.co/api/models/GoToCompany/Llama-Sahabat-AI-v2-70B-IT", "https://developingtelecoms.com/telecom-technology/enterprise-ecosystems/18569-massive-enhancement-announced-for-indonesian-llm.html", "https://www.gotocompany.com/en/news/press/sahabat-ai-gets-smarter-indosat-and-goto-launch-new-70-billion-parameter-model-with-multilingual-chat-service" ], "unverified": [ "indonesia_hosting", "indonesian_measured" ], "indonesian_measured": [ { "benchmark": "IndoMMLU and SEA-HELM (model card says both were run)", "score": null, "metric": null, "run_by": "GoTo / Sahabat-AI team (developer-reported)", "url": "https://huggingface.co/GoToCompany/Llama-Sahabat-AI-v2-70B-IT" } ] }, { "id": "goto-llama3-8b-cpt-sahabatai-v1", "name": "llama3-8b-cpt-sahabatai-v1", "level": "model", "family_id": "goto-sahabat-ai-v1", "developer": [ "goto", "indosat", "ai-singapore" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-11-06", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "llama-3-community", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": "Open weights; on-prem or Indonesian territory; sovereign cloud GPU Merdeka by Lintasarta", "runtimes": [], "languages_note": "Indonesian + regional languages", "indonesian_support": "listed by developer", "regional_languages": [ "Javanese", "Sundanese", "Batak Toba", "Balinese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Sahabat-AI v1 continued-pretraining model (base and instruct).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/GoToCompany", "https://huggingface.co/models?search=sahabat", "https://blogs.nvidia.com/?p=75648" ], "unverified": [ "licence", "indonesian_measured", "indonesia_hosting" ] }, { "id": "goto-sahabat-ai", "name": "Sahabat-AI", "level": "family", "family_id": null, "developer": [ "goto", "ai-singapore" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "listed by developer", "regional_languages": [ "Javanese", "Sundanese", "Batak Toba", "Balinese" ], "indonesia_hosting": "unknown", "status": null, "summary": "Indonesian-language LLM family from GoTo and AI Singapore, co-initiated by Indosat; launched Nov 2024 on a Llama 3 8B fine-tune with NVIDIA NeMo. Used in Indosat's citizen chatbot and GoTo's Dira voice assistant.", "verified_on": "2026-10-10", "sources": [ "https://blogs.nvidia.com/?p=75648", "https://huggingface.co/GoToCompany/Llama-Sahabat-AI-v2-70B-IT", "https://huggingface.co/GoToCompany", "https://developingtelecoms.com/telecom-technology/enterprise-ecosystems/18569-massive-enhancement-announced-for-indonesian-llm.html" ], "unverified": [ "indonesia_hosting", "indonesian_measured" ], "indonesian_measured": [ { "benchmark": "IndoMMLU and SEA-HELM (model card says both were run)", "score": null, "metric": null, "run_by": "GoTo / Sahabat-AI team (developer-reported)", "url": "https://huggingface.co/GoToCompany/Llama-Sahabat-AI-v2-70B-IT" } ] }, { "id": "goto-sahabat-ai-v1", "name": "Sahabat-AI v1", "level": "family", "family_id": null, "developer": [ "goto", "indosat" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-11-06", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Indonesian + regional languages", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "GoTo and Indosat's first Sahabat-AI models, continued pretraining on Gemma 2 9B and Llama 3 8B (\"SEALION-based foundations\").", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/GoToCompany", "https://huggingface.co/models?search=sahabat" ], "unverified": [ "licence", "indonesian_measured", "indonesia_hosting" ] }, { "id": "grab-custom-vision-llm-1b", "name": "Grab custom vision LLM 1B", "level": "model", "family_id": null, "developer": [ "grab" ], "model_class": "OCR and document", "modalities_in": [ "image", "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Qwen2-VL 2B vision encoder plus Qwen2.5 0.5B decoder", "params_total": 1000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [ "internal only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Bahasa Indonesia, Thai, Vietnamese, English", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Grab's eKYC document extraction vision LLM serving live production traffic (post 4 Nov 2025); within about 3 pts of the 2B model.", "verified_on": "2026-10-10", "sources": [ "https://engineering.grab.com/custom-vision-llm-at-grab" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "grab-custom-vision-llm-2b", "name": "Grab custom vision LLM 2B", "level": "model", "family_id": null, "developer": [ "grab" ], "model_class": "OCR and document", "modalities_in": [ "image", "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 2000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [ "internal only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Bahasa Indonesia, Thai, Vietnamese, English", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Earlier 2B Grab eKYC vision LLM (full fine-tune): Thai +70 pts, Vietnamese +40 pts vs baseline.", "verified_on": "2026-10-10", "sources": [ "https://engineering.grab.com/custom-vision-llm-at-grab" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "guanaco", "name": "Guanaco", "level": "model", "family_id": null, "developer": [ "university-of-washington" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-05-23", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "QLoRA-trained model reaching 99.3% of ChatGPT on the Vicuna benchmark after 24 hours on one GPU.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2305.14314", "https://arxiv.org/html/2305.14314" ], "unverified": [] }, { "id": "gunadarma-university-yolov8-ffb-grader", "name": "YOLOv8 FFB ripeness grader (Gunadarma)", "level": "model", "family_id": null, "developer": [ "gunadarma-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-12-31", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "research only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Academic FFB grading model in a Streamlit web app, trained on 6,592 Roboflow images; precision 94.5%, recall 94.7%, mAP 98%. No plantation deployment stated.", "verified_on": "2026-10-10", "sources": [ "https://ejournal.gunadarma.ac.id/index.php/tekno/article/view/67" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "harvard-university-nequip-oam-l", "name": "Nequip-OAM-L", "level": "model", "family_id": null, "developer": [ "harvard-university" ], "model_class": "materials", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "GNN", "architecture_detail": "Equivariant GNN interatomic potential (MLIP)", "params_total": 9600000, "params_active": null, "params_checkpoint": null, "params_note": "Training data OMat24+sAlex+MPtrj", "context_length": null, "release_date": "2025-09-08", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Machine-learned interatomic potential on Matbench Discovery (rank 7 by CPS, date added 2025-09-08).", "verified_on": "2026-10-10", "sources": [ "https://matbench-discovery.materialsproject.org/" ], "unverified": [ "developer" ] }, { "id": "harvard-university-nequip-oam-xl", "name": "Nequip-OAM-XL", "level": "model", "family_id": null, "developer": [ "harvard-university" ], "model_class": "materials", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "GNN", "architecture_detail": "Equivariant GNN interatomic potential (MLIP)", "params_total": 32100000, "params_active": null, "params_checkpoint": null, "params_note": "Training data OMat24+MPtrj+sAlex", "context_length": null, "release_date": "2025-11-30", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Machine-learned interatomic potential on Matbench Discovery (rank 4 by CPS, date added 2025-11-30).", "verified_on": "2026-10-10", "sources": [ "https://matbench-discovery.materialsproject.org/" ], "unverified": [ "developer" ] }, { "id": "hexgrad-kokoro-82m", "name": "Kokoro-82M", "level": "model", "family_id": null, "developer": [ "hexgrad" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": "StyleTTS 2 plus ISTFTNet", "params_total": 82000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-01-27", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge / CPU class", "runtimes": [], "languages_note": "8 languages (v1.0); Indonesian not stated", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tiny open TTS model with 11.2M Hugging Face downloads.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/hexgrad/Kokoro-82M" ], "unverified": [] }, { "id": "hexgrad-kokoro-v0-19", "name": "Kokoro v0.19", "level": "model", "family_id": null, "developer": [ "hexgrad" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier Kokoro version.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/hexgrad/Kokoro-82M" ], "unverified": [] }, { "id": "heygen-avatar-iv", "name": "HeyGen Avatar IV", "level": "model", "family_id": null, "developer": [ "heygen" ], "model_class": "avatar", "modalities_in": [ "text", "image" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "HeyGen avatar SaaS with a verbal consent check; reported misuse in deepfake scams.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/HeyGen" ], "unverified": [] }, { "id": "hidream-ai-hidream", "name": "HiDream", "level": "family", "family_id": null, "developer": [ "hidream-ai" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "HiDream.ai image and video models (I1, O1).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/HiDream-ai/HiDream-I1-Full" ], "unverified": [] }, { "id": "hidream-ai-hidream-i1", "name": "HiDream-I1", "level": "model", "family_id": "hidream-ai-hidream", "developer": [ "hidream-ai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": "Sparse DiT; T5-XXL plus Llama 3.1 8B encoders; FLUX.1 schnell VAE", "params_total": 17000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-05-28", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "HiDream.ai 17B sparse DiT image model; transformer under MIT.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/HiDream-ai/HiDream-I1-Full", "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text" ], "unverified": [ "deployment_tiers" ] }, { "id": "hidream-ai-hidream-i1-dev", "name": "HiDream-I1-Dev", "level": "variant", "family_id": "hidream-ai-hidream", "developer": [ "hidream-ai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-04-06", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Dev variant of HiDream-I1 on Artificial Analysis.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text", "https://huggingface.co/HiDream-ai/HiDream-I1-Dev" ], "unverified": [] }, { "id": "hidream-ai-hidream-o1-image", "name": "HiDream-O1-Image", "level": "model", "family_id": "hidream-ai-hidream", "developer": [ "hidream-ai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "HiDream O1 image model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/HiDream-ai/HiDream-I1-Full" ], "unverified": [] }, { "id": "hidream-ai-hidream-o1-image-1-5", "name": "HiDream-O1-Image-1.5", "level": "model", "family_id": "hidream-ai-hidream", "developer": [ "hidream-ai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "HiDream O1 image model version 1.5, open weights on Artificial Analysis board.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text", "https://huggingface.co/HiDream-ai/HiDream-I1-Full" ], "unverified": [] }, { "id": "hidream-ai-hidream-o1-video", "name": "HiDream-O1-Video", "level": "model", "family_id": null, "developer": [ "hidream-ai" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "HiDream.ai video model leading VBench (v1) text-to-video in Aug 2026.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/spaces/Vchitect/VBench_Leaderboard" ], "unverified": [] }, { "id": "huawei-pangu-5-5", "name": "Pangu 5.5", "level": "family", "family_id": null, "developer": [ "huawei" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-06", "release_date_basis": "notes", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": "Huawei Cloud ModelArts, hybrid cloud; mining companies fine-tune on ModelArts.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Huawei's industrial model family (NLP, CV, Prediction, Multimodal and World Model) used in mining, energy, oil and cement in China.", "verified_on": "2026-10-10", "sources": [ "https://aimagazine.com/articles/how-huawei-pangu-5-5-ai-models-transform-industry-operations", "https://www.intelligentcio.com/apac/2025/06/24/cross-industry-use-cases-for-huaweis-pangu-models-5-5-showcased/", "https://www.mining-technology.com/newsletters/huaweis-pangu-5-5-ai-intelligent-mining-in-action" ], "unverified": [] }, { "id": "huawei-pangu-5-5-nlp", "name": "Pangu 5.5 NLP", "level": "model", "family_id": "huawei-pangu-5-5", "developer": [ "huawei" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": "MoE", "architecture_detail": "256 experts", "params_total": 718000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "718B MoE language model in the Pangu 5.5 family.", "verified_on": "2026-10-10", "sources": [ "https://aimagazine.com/articles/how-huawei-pangu-5-5-ai-models-transform-industry-operations" ], "unverified": [] }, { "id": "huawei-pangu-cv", "name": "Pangu CV", "level": "model", "family_id": "huawei-pangu-5-5", "developer": [ "huawei" ], "model_class": "vision", "modalities_in": [ "image", "video" ], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": "Cloud plus edge cameras", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Huawei vision model for mines (cracks, belt damage, hot bearings, leaks, loose rock); won Mining Technology Excellence Awards 2025.", "verified_on": "2026-10-10", "sources": [ "https://www.intelligentcio.com/apac/2025/06/24/cross-industry-use-cases-for-huaweis-pangu-models-5-5-showcased/", "https://www.mining-technology.com/newsletters/huaweis-pangu-5-5-ai-intelligent-mining-in-action" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "huawei-pangu-mine-model", "name": "Pangu Mine Model", "level": "model", "family_id": null, "developer": [ "huawei" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023", "release_date_basis": "notes", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Huawei mining model from 2023, used at Shandong Energy.", "verified_on": "2026-10-10", "sources": [ "https://www.mining-technology.com/newsletters/huaweis-pangu-5-5-ai-intelligent-mining-in-action" ], "unverified": [ "indonesian_measured", "indonesia_hosting", "model_class" ] }, { "id": "huawei-pangu-multimodal-world-model", "name": "Pangu Multimodal and World Model", "level": "model", "family_id": "huawei-pangu-5-5", "developer": [ "huawei" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Huawei multimodal and world model used for autonomous haulage at the Yimin open-pit coal mine (100 autonomous electric trucks, May 2025).", "verified_on": "2026-10-10", "sources": [ "https://aimagazine.com/articles/how-huawei-pangu-5-5-ai-models-transform-industry-operations", "https://www.mining-technology.com/newsletters/huaweis-pangu-5-5-ai-intelligent-mining-in-action" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "huawei-pangu-prediction-model", "name": "Pangu Prediction Model", "level": "model", "family_id": "huawei-pangu-5-5", "developer": [ "huawei" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Size not disclosed", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Huawei forecasting model used for Conch Cement clinker strength (3- and 28-day) and Shenzhen Energy wind and solar yield.", "verified_on": "2026-10-10", "sources": [ "https://aimagazine.com/articles/how-huawei-pangu-5-5-ai-models-transform-industry-operations" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "huawei-pangu-weather", "name": "Pangu-Weather", "level": "model", "family_id": null, "developer": [ "huawei" ], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "other", "architecture_detail": "3D high-resolution neural network; 0.25 deg (721x1440)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-07-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-sa-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Huawei Cloud first-generation AI weather model; commercial use forbidden.", "verified_on": "2026-10-10", "sources": [ "https://github.com/198808xc/Pangu-Weather", "https://news.tuoitre.vn/china-bets-on-ai-weather-forecasting-as-extreme-weather-intensifies-103260812213327362.htm" ], "unverified": [] }, { "id": "hugging-face-smollm2", "name": "SmolLM2", "level": "family", "family_id": null, "developer": [ "hugging-face" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "135M to 1.7B", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "CPU server", "laptop", "mobile or NPU" ], "footprint_note": "Ollama GGUF builds", "runtimes": [ "ollama" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Hugging Face small model family in the Ollama library.", "verified_on": "2026-10-10", "sources": [ "https://ollama.com/library" ], "unverified": [] }, { "id": "hugging-face-smollm3", "name": "SmolLM3", "level": "model", "family_id": null, "developer": [ "hugging-face" ], "model_class": "edge LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": 3000000000, "params_active": null, "params_checkpoint": null, "params_note": "About 11T tokens; 384 H100 for 24 days; fully open recipe.", "context_length": null, "release_date": "2025-07-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "fully open", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Hugging Face 3B small model with a fully open training recipe.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/blog/smollm3" ], "unverified": [] }, { "id": "hugging-face-smollm3-3b", "name": "SmolLM3-3B", "level": "model", "family_id": null, "developer": [ "hugging-face" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "Decoder-only with GQA and NoPE (3:1); dual-mode reasoning (/think, /no_think); pretrained from scratch on 11.2T tokens, then 140B reasoning tokens of midtraining, SFT and APO; trained on 384 H100", "params_total": 3000000000, "params_active": null, "params_checkpoint": null, "params_note": "Trained at 64K context, 128K via YaRN", "context_length": 128000, "release_date": "2025-07-08", "release_date_basis": "notes", "weights_date": null, "openness": "fully open", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "laptop", "mobile or NPU", "CPU server" ], "footprint_note": "Quantized footprint unverified", "runtimes": [], "languages_note": "6 languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Hugging Face's 3B small model, historically fully open; no SmolLM4 found.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/HuggingFaceTB", "https://huggingface.co/HuggingFaceTB/SmolLM3-3B", "https://huggingface.co/api/models/HuggingFaceTB/SmolLM3-3B", "https://osai-index.eu/model/smollm", "https://huggingface.co/blog/smollm3", "https://huggingface.co/datasets/HuggingFaceTB/smoltalk2" ], "unverified": [ "footprint_note" ] }, { "id": "hugging-face-smollm3-3b-base", "name": "SmolLM3-3B-Base", "level": "model", "family_id": null, "developer": [ "hugging-face" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "3B", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Base checkpoint of SmolLM3 3B.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/HuggingFaceTB/SmolLM3-3B", "https://huggingface.co/api/models/HuggingFaceTB/SmolLM3-3B" ], "unverified": [] }, { "id": "ibm-flowstate-r1-1", "name": "FlowState r1.1", "level": "model", "family_id": null, "developer": [ "ibm" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": null, "params_total": 9070000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-09-10", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "IBM TSFM compact forecasting model.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2605.20119", "https://huggingface.co/spaces/Salesforce/GIFT-Eval", "https://huggingface.co/api/models/ibm-granite/granite-timeseries-flowstate-r1" ], "unverified": [] }, { "id": "ibm-granite-4-0", "name": "Granite 4.0", "level": "family", "family_id": null, "developer": [ "ibm" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "IBM Granite 4.0 family of hybrid Mamba and transformer models (H-Small, H-Tiny, H-Micro) plus a conventional Micro model, released under Apache 2.0.", "verified_on": "2026-10-11", "sources": [ "https://www.ibm.com/new/announcements/ibm-granite-4-0-hyper-efficient-high-performance-hybrid-models" ], "unverified": [] }, { "id": "ibm-granite-4-0-1b-speech", "name": "granite-4.0-1b-speech", "level": "model", "family_id": null, "developer": [ "ibm" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "autoregressive", "architecture_detail": null, "params_total": 2313207132, "params_active": null, "params_checkpoint": 2310000000, "params_note": "1B by name; HF reports 2.31B total, conflicts with name (unverified)", "context_length": null, "release_date": "2026-02-27", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "IBM Granite 4.0 speech recognition model.", "verified_on": "2026-10-10", "sources": [ "https://benchmarklist.com/benchmarks/open_asr_leaderboard/", "https://huggingface.co/api/models/ibm-granite/granite-4.0-1b-speech" ], "unverified": [] }, { "id": "ibm-granite-4-0-h-micro", "name": "Granite-4.0-H-Micro", "level": "model", "family_id": "ibm-granite-4-0", "developer": [ "ibm" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Dense hybrid: 4 attention layers and 36 Mamba2 layers, GQA, SwiGLU MLP, NoPE", "params_total": 3000000000, "params_active": 3000000000, "params_checkpoint": null, "params_note": null, "context_length": 128000, "release_date": "2025-10-02", "release_date_basis": "announcement page", "weights_date": "2025-10-02", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": "IBM says the Tiny and Micro models are designed for low latency, edge and local applications. OpenRouter lists this checkpoint as Granite 4.0 Micro", "runtimes": [], "languages_note": "English, German, Spanish, French, Japanese, Portuguese, Arabic, Czech, Italian, Korean, Dutch, Chinese", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "IBM 3B long-context instruct model with a hybrid Mamba2 and attention design, fine-tuned for instruction following and tool calling.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/ibm-granite/granite-4.0-h-micro", "https://www.ibm.com/new/announcements/ibm-granite-4-0-hyper-efficient-high-performance-hybrid-models", "https://openrouter.ai/ibm-granite/granite-4.0-h-micro" ], "unverified": [] }, { "id": "ibm-granite-4-0-h-micro-base", "name": "Granite-4.0-H-Micro-Base", "level": "model", "family_id": "ibm-granite-4-0", "developer": [ "ibm" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Base model from which Granite-4.0-H-Micro is fine-tuned.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/ibm-granite/granite-4.0-h-micro" ], "unverified": [] }, { "id": "ibm-granite-4-1-8b", "name": "granite-4.1-8b", "level": "model", "family_id": null, "developer": [ "ibm" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/ibm-granite/granite-guardian-4.1-8b" ], "unverified": [] }, { "id": "ibm-granite-4-1-8b-base", "name": "Granite-4.1-8B-Base", "level": "model", "family_id": null, "developer": [ "ibm" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/ibm-granite/granite-4.2-8b" ], "unverified": [] }, { "id": "ibm-granite-4-2", "name": "Granite 4.2", "level": "family", "family_id": null, "developer": [ "ibm" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "3B, 8B, 30B sizes", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU", "laptop" ], "footprint_note": "T4/T5 estimate (unverified). Updated Sep 2026.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "IBM general Granite 4.2 family; coding folded in, no separate Granite Code 4.x models listed.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/ibm-granite", "https://huggingface.co/ibm-granite/granite-4.2-8b" ], "unverified": [ "deployment_tiers" ] }, { "id": "ibm-granite-4-2-30b", "name": "Granite 4.2 30B", "level": "model", "family_id": "ibm-granite-4-2", "developer": [ "ibm" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "Dense transformer, GQA (not Mamba)", "params_total": 30000000000, "params_active": null, "params_checkpoint": null, "params_note": "128K native, extends to 512K", "context_length": 128000, "release_date": "2026-09", "release_date_basis": "registry date", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": "FP8 and GGUF variants", "runtimes": [], "languages_note": "12 languages listed; no Indonesian", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "IBM Granite 4.2 dense text model (30B), weights and code with data partly documented.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/ibm-granite", "https://huggingface.co/ibm-granite/granite-4.2-8b" ], "unverified": [] }, { "id": "ibm-granite-4-2-3b", "name": "Granite 4.2 3B", "level": "model", "family_id": "ibm-granite-4-2", "developer": [ "ibm" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "Dense transformer, GQA (not Mamba)", "params_total": 3000000000, "params_active": null, "params_checkpoint": null, "params_note": "128K native, extends to 512K", "context_length": 128000, "release_date": "2026-09", "release_date_basis": "registry date", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "FP8 and GGUF variants", "runtimes": [], "languages_note": "12 languages listed; no Indonesian", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "IBM Granite 4.2 dense text model (3B), weights and code with data partly documented.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/ibm-granite", "https://huggingface.co/ibm-granite/granite-4.2-8b" ], "unverified": [] }, { "id": "ibm-granite-4-2-8b", "name": "Granite 4.2 8B", "level": "model", "family_id": "ibm-granite-4-2", "developer": [ "ibm" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "Dense transformer, GQA (not Mamba)", "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": "128K native, extends to 512K", "context_length": 128000, "release_date": "2026-09", "release_date_basis": "registry date", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": "FP8 and GGUF variants", "runtimes": [], "languages_note": "12 languages listed; no Indonesian", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "IBM Granite 4.2 dense text model (8B), weights and code with data partly documented.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/ibm-granite", "https://huggingface.co/ibm-granite/granite-4.2-8b" ], "unverified": [] }, { "id": "ibm-granite-4-350m", "name": "Granite 4 350M", "level": "model", "family_id": null, "developer": [ "ibm" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "350M", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "CPU server", "laptop", "mobile or NPU" ], "footprint_note": "Ollama GGUF build", "runtimes": [ "ollama" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small model listed in the Ollama library for CPU, laptop and phone use.", "verified_on": "2026-10-10", "sources": [ "https://ollama.com/library" ], "unverified": [] }, { "id": "ibm-granite-embedding-multilingual-r2", "name": "Granite Embedding Multilingual R2", "level": "model", "family_id": null, "developer": [ "ibm" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-05-14", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "IBM's multilingual embedding model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/ibm-granite" ], "unverified": [ "params_total", "licence" ] }, { "id": "ibm-granite-flowstate", "name": "Granite FlowState", "level": "model", "family_id": null, "developer": [ "ibm" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": null, "architecture_detail": "Sampling-rate-agnostic forecasting", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "'few million' parameters", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "CPU server" ], "footprint_note": "GPU-free", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "IBM forecasting model, #3 on GIFT-Eval as of 2 Oct 2025 per IBM docs. Licence unverified.", "verified_on": "2026-10-10", "sources": [ "https://www.ibm.com/ae-ar/granite/docs/models/time-series/" ], "unverified": [ "licence" ] }, { "id": "ibm-granite-geospatial-ocean", "name": "Granite-Geospatial-Ocean", "level": "model", "family_id": null, "developer": [ "ibm" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Derivative of Prithvi-EO.", "verified_on": "2026-10-10", "sources": [ "https://research.ibm.com/blog/earth-observation-leadership" ], "unverified": [] }, { "id": "ibm-granite-guardian-3-3-8b", "name": "Granite Guardian 3.3-8B", "level": "model", "family_id": null, "developer": [ "ibm" ], "model_class": "guard", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "IBM Granite Guardian 3.3-8B guard model (2024 to Apr 2026 range).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/ibm-granite/models?search=guardian" ], "unverified": [] }, { "id": "ibm-granite-guardian-4-0-3b-toxicity-ja", "name": "Granite Guardian 4.0-3B toxicity-ja", "level": "model", "family_id": null, "developer": [ "ibm" ], "model_class": "guard", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 3000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Japanese", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "IBM Granite Guardian 4.0-3B toxicity-ja guard model (2024 to Apr 2026 range).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/ibm-granite/models?search=guardian" ], "unverified": [] }, { "id": "ibm-granite-guardian-4-1-8b", "name": "Granite Guardian 4.1-8B", "level": "model", "family_id": null, "developer": [ "ibm" ], "model_class": "guard", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English only", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "IBM's guard model covering harm (bias, jailbreak, violence, profanity, unethical behaviour), RAG groundedness and relevance, and function-calling hallucination.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/ibm-granite/granite-guardian-4.1-8b", "https://huggingface.co/ibm-granite/models?search=guardian" ], "unverified": [] }, { "id": "ibm-granite-guardian-hap-125m", "name": "Granite Guardian HAP 125M", "level": "model", "family_id": null, "developer": [ "ibm" ], "model_class": "guard", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 125000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "IBM Granite Guardian HAP 125M guard model (2024 to Apr 2026 range).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/ibm-granite/models?search=guardian" ], "unverified": [] }, { "id": "ibm-granite-guardian-hap-38m", "name": "Granite Guardian HAP 38M", "level": "model", "family_id": null, "developer": [ "ibm" ], "model_class": "guard", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 38000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "IBM Granite Guardian HAP 38M guard model (2024 to Apr 2026 range).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/ibm-granite/models?search=guardian" ], "unverified": [] }, { "id": "ibm-granite-patchtst-fm-r1", "name": "Granite PatchTST-FM r1", "level": "model", "family_id": null, "developer": [ "ibm", "rensselaer-polytechnic-institute" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "dense transformer", "architecture_detail": "PatchTST lineage", "params_total": 257900000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-03-11", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "IBM TSFM and RPI PatchTST foundation model.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2605.20119", "https://huggingface.co/spaces/Salesforce/GIFT-Eval", "https://huggingface.co/api/models/ibm-granite/granite-timeseries-patchtst-fm-r1" ], "unverified": [] }, { "id": "ibm-granite-patchtst-fm-r2", "name": "Granite PatchTST-FM r2", "level": "model", "family_id": null, "developer": [ "ibm", "rensselaer-polytechnic-institute" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "r2 of PatchTST-FM, now on GIFT-Eval.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/spaces/Salesforce/GIFT-Eval" ], "unverified": [] }, { "id": "ibm-granite-speech-3-3-8b", "name": "granite-speech-3.3-8b", "level": "model", "family_id": null, "developer": [ "ibm" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "autoregressive", "architecture_detail": "Speech encoder + Granite LLM", "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "IBM Granite Speech 3.3 ASR model.", "verified_on": "2026-10-10", "sources": [ "https://www.gladia.io/blog/best-open-source-speech-to-text-models", "https://benchmarklist.com/benchmarks/open_asr_leaderboard/" ], "unverified": [] }, { "id": "ibm-granite-speech-4-1-2b", "name": "granite-speech-4.1-2b", "level": "model", "family_id": null, "developer": [ "ibm" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "autoregressive", "architecture_detail": "Conformer encoder + LLM decoder (top-4 pattern)", "params_total": 2310000000, "params_active": null, "params_checkpoint": null, "params_note": "2B by name; HF reports 2.31B", "context_length": null, "release_date": "2026-04-16", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "IBM Granite speech recognition model; rank 1 on the Open ASR Leaderboard snapshot of 6 May 2026.", "verified_on": "2026-10-10", "sources": [ "https://benchmarklist.com/benchmarks/open_asr_leaderboard/", "https://huggingface.co/api/models/ibm-granite/granite-speech-4.1-2b" ], "unverified": [] }, { "id": "ibm-granite-tspulse", "name": "Granite TSPulse", "level": "model", "family_id": null, "developer": [ "ibm" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": null, "architecture_detail": "Dual time and frequency-space pretraining", "params_total": 1080000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-05-19", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "CPU, GPU-free", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "IBM tiny time-series model for anomaly detection and more (paper 19 May 2025, ICLR 2026). Claims #1 on TSB-AD univariate and multivariate (+20%) with no scores on the card. Maximo integration unverified.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/ibm-granite/granite-timeseries-tspulse-r1", "https://alphasignal.ai/news/ibm-s-tiny-tspulse-beats-models-100x-bigger-across-75-datasets" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "ibm-granite-ttm-r1", "name": "Granite TTM-R1", "level": "model", "family_id": null, "developer": [ "ibm" ], "model_class": "time series", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Predecessor of TTM-R2; r2 is reported more than 15% better.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/ibm-granite/granite-timeseries-ttm-r2" ], "unverified": [] }, { "id": "ibm-granite-ttm-r2", "name": "Granite TinyTimeMixer (TTM) r2", "level": "model", "family_id": null, "developer": [ "ibm" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": "TinyTimeMixer (MLP-mixer family)", "params_total": 805000, "params_active": null, "params_checkpoint": null, "params_note": "about 805k to 1M+", "context_length": null, "release_date": "2024", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [ "CPU server", "laptop" ], "footprint_note": "CPU-only machines and laptops; watsonx.ai forecast API.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "IBM tiny forecasting model for CPU-only machines; vendor claims to beat larger models, but GIFT-Eval ranks TTM-R2 140 of 143 with test leakage flagged.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/ibm-granite/granite-timeseries-ttm-r2", "https://huggingface.co/spaces/Salesforce/GIFT-Eval", "https://www.ibm.com/ae-ar/granite/docs/models/time-series/" ], "unverified": [ "release_date", "architecture_detail" ] }, { "id": "ibm-granite-ttm-r2-1", "name": "Granite TinyTimeMixer (TTM) r2.1", "level": "model", "family_id": null, "developer": [ "ibm" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server", "laptop" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "TTM r2.1 adds daily/weekly data.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/ibm-granite/granite-timeseries-ttm-r2" ], "unverified": [] }, { "id": "ibm-prithvi-eo-1-0", "name": "Prithvi-EO 1.0", "level": "model", "family_id": null, "developer": [ "ibm", "nasa" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First Prithvi EO model.", "verified_on": "2026-10-10", "sources": [ "https://research.ibm.com/blog/earth-observation-leadership" ], "unverified": [] }, { "id": "ibm-prithvi-eo-2-0", "name": "Prithvi-EO-2.0", "level": "family", "family_id": null, "developer": [ "ibm", "nasa", "forschungszentrum-julich" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": "MAE ViT, multi-temporal; TL = temporal + location embeddings", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "tiny-TL 5M, 100M-TL, 300M(-TL), 600M(-TL)", "context_length": null, "release_date": "2024-12-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "embedded GPU", "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "IBM and NASA Earth observation foundation model; 600M-TL is 8% above Prithvi 1.0 on GEO-Bench and beats six other GFMs; leads EO-specific categories on GEO-Bench-2.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2412.02732v2", "https://huggingface.co/ibm-nasa-geospatial/Prithvi-EO-2.0-600M", "https://research.ibm.com/blog/earth-observation-leadership", "https://github.com/NASA-IMPACT/Prithvi-EO-2.0" ], "unverified": [] }, { "id": "ibm-prithvi-eo-2-0-100m-tl", "name": "Prithvi-EO-2.0 100M-TL", "level": "variant", "family_id": "ibm-prithvi-eo-2-0", "developer": [ "ibm", "nasa" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 100000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-12-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Prithvi-EO-2.0 size variant.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/ibm-nasa-geospatial/Prithvi-EO-2.0-600M", "https://research.ibm.com/blog/earth-observation-leadership" ], "unverified": [] }, { "id": "ibm-prithvi-eo-2-0-300m", "name": "Prithvi-EO-2.0 300M", "level": "variant", "family_id": "ibm-prithvi-eo-2-0", "developer": [ "ibm", "nasa" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": "ViT masked autoencoder pretrained from scratch on NASA HLS V2, 30 m, about 4.2M samples, 6 bands", "params_total": 300000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-12-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [ "terratorch" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Prithvi-EO-2.0 size variant.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/ibm-nasa-geospatial/Prithvi-EO-2.0-600M", "https://research.ibm.com/blog/earth-observation-leadership", "https://arxiv.org/abs/2412.02732v2", "https://github.com/NASA-IMPACT/Prithvi-EO-2.0" ], "unverified": [] }, { "id": "ibm-prithvi-eo-2-0-600m", "name": "Prithvi-EO-2.0 600M", "level": "variant", "family_id": "ibm-prithvi-eo-2-0", "developer": [ "ibm", "nasa" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": "ViT masked autoencoder pretrained from scratch on NASA HLS V2, 30 m, about 4.2M samples, 6 bands", "params_total": 600000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-12-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [ "terratorch" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Prithvi-EO-2.0 size variant.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/ibm-nasa-geospatial/Prithvi-EO-2.0-600M", "https://research.ibm.com/blog/earth-observation-leadership", "https://arxiv.org/abs/2412.02732v2", "https://github.com/NASA-IMPACT/Prithvi-EO-2.0" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "ibm-prithvi-eo-2-0-600m-tl", "name": "Prithvi-EO-2.0-600M-TL", "level": "variant", "family_id": "ibm-prithvi-eo-2-0", "developer": [ "ibm", "nasa" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": "ViT masked autoencoder pretrained from scratch on NASA HLS V2, 30 m, about 4.2M samples, 6 bands", "params_total": 600000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-12-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [ "terratorch" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "TL variant, 8% above Prithvi v1 across GEO-Bench tasks; beats six other GFMs at 0.1 to 15 m.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/ibm-nasa-geospatial/Prithvi-EO-2.0-600M", "https://arxiv.org/abs/2412.02732v2", "https://arxiv.org/abs/2412.02732" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "ibm-prithvi-eo-2-0-tiny-tl", "name": "Prithvi-EO-2.0 TINY-TL", "level": "variant", "family_id": "ibm-prithvi-eo-2-0", "developer": [ "ibm", "nasa" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 5000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-12-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "embedded GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Prithvi-EO-2.0 size variant.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/ibm-nasa-geospatial/Prithvi-EO-2.0-600M", "https://research.ibm.com/blog/earth-observation-leadership" ], "unverified": [] }, { "id": "ibm-prithvi-v1", "name": "Prithvi v1", "level": "model", "family_id": null, "developer": [ "ibm", "nasa" ], "model_class": "geospatial", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Predecessor of Prithvi-EO-2.0.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2412.02732v2" ], "unverified": [] }, { "id": "ibm-prithvi-wxc", "name": "Prithvi-WxC", "level": "model", "family_id": null, "developer": [ "ibm", "nasa" ], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 2300000000, "params_active": null, "params_checkpoint": null, "params_note": "2.3B per repo name", "context_length": null, "release_date": "2024-09-20", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "cdla-permissive-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "IBM and NASA weather and climate foundation model.", "verified_on": "2026-10-10", "sources": [ "https://research.ibm.com/blog/earth-observation-leadership", "https://huggingface.co/api/models/ibm-nasa-geospatial/Prithvi-WxC-1.0-2300M" ], "unverified": [] }, { "id": "ibm-surya", "name": "Surya", "level": "model", "family_id": null, "developer": [ "ibm", "nasa" ], "model_class": "scientific", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-08-13", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "IBM, NASA and eight centres solar physics foundation model with SuryaBench.", "verified_on": "2026-10-10", "sources": [ "https://research.ibm.com/blog/earth-observation-leadership", "https://huggingface.co/api/models/nasa-ibm-ai4science/Surya-1.0" ], "unverified": [] }, { "id": "ibm-terracodec", "name": "TerraCodec", "level": "model", "family_id": null, "developer": [ "ibm" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "other", "architecture_detail": "Neural compression for Sentinel-2", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "1M to 10M", "context_length": null, "release_date": "2025", "release_date_basis": "notes", "weights_date": null, "openness": "open source", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "embedded GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Neural image compression for Sentinel-2; up to 10x better than JPEG 2000/HEVC at equal quality (vendor).", "verified_on": "2026-10-10", "sources": [ "https://research.ibm.com/blog/earth-observation-leadership" ], "unverified": [ "release_date" ] }, { "id": "ibm-terramind", "name": "TerraMind", "level": "model", "family_id": null, "developer": [ "ibm", "esa", "forschungszentrum-julich" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "other", "architecture_detail": "Any-to-any multimodal generative EO model, Thinking in Modalities; trained on TerraMesh", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "base, small, tiny", "context_length": null, "release_date": "2025-02-14", "release_date_basis": "registry date", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "embedded GPU", "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "IBM, ESA and Julich generative EO model (ICCV 2025); leads PANGAEA and three GEO-Bench-2 capability areas (vendor).", "verified_on": "2026-10-10", "sources": [ "https://research.ibm.com/blog/earth-observation-leadership", "https://huggingface.co/api/models/ibm-esa-geospatial/TerraMind-1.0-base" ], "unverified": [] }, { "id": "icams-bochum-grace-2l-oam-l", "name": "GRACE-2L-OAM-L", "level": "model", "family_id": null, "developer": [ "icams-bochum" ], "model_class": "materials", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "GNN", "architecture_detail": "Equivariant GNN interatomic potential (MLIP)", "params_total": 26400000, "params_active": null, "params_checkpoint": null, "params_note": "Training data OMat24+sAlex+MPtrj", "context_length": null, "release_date": "2025-09-09", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Machine-learned interatomic potential on Matbench Discovery (rank 9 by CPS, date added 2025-09-09).", "verified_on": "2026-10-10", "sources": [ "https://matbench-discovery.materialsproject.org/" ], "unverified": [] }, { "id": "ideogram", "name": "Ideogram", "level": "family", "family_id": null, "developer": [ "ideogram" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Ideogram image model line (Ideogram 1 to 4).", "verified_on": "2026-10-10", "sources": [ "https://ideogram.ai/news/ideogram-4.0/" ], "unverified": [] }, { "id": "ideogram-3", "name": "Ideogram 3", "level": "model", "family_id": "ideogram", "developer": [ "ideogram" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier Ideogram image model.", "verified_on": "2026-10-10", "sources": [ "https://ideogram.ai/news/ideogram-4.0/" ], "unverified": [] }, { "id": "ideogram-4-0", "name": "Ideogram 4.0", "level": "model", "family_id": "ideogram", "developer": [ "ideogram" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": 9300000000, "params_active": null, "params_checkpoint": null, "params_note": "9.3B per a third-party title (unverified)", "context_length": null, "release_date": "2026-06-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "ideogram-open-model", "access_routes": [ "open weights download", "API", "consumer app" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open-weight image model under the commercial Ideogram Open Model licence, with multilingual text rendering, bounding-box layout and brand fine-tuning.", "verified_on": "2026-10-10", "sources": [ "https://ideogram.ai/news/ideogram-4.0/", "https://arena.ai/leaderboard/text-to-image", "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text" ], "unverified": [ "params_total", "deployment_tiers" ] }, { "id": "ideogram-v2", "name": "Ideogram v2", "level": "model", "family_id": "ideogram", "developer": [ "ideogram" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-08-21", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "consumer app" ], "deployment_tiers": [], "footprint_note": "API model value V_2 (default in the legacy generate endpoint) and V_2_TURBO", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Ideogram 2.0 is an Ideogram text-to-image model with General, Realistic, Design, 3D and Anime styles, offered free on the web, in an iOS app and through a beta API at launch.", "verified_on": "2026-10-11", "sources": [ "https://tech.yahoo.com/computing/articles/ideogram-2-0-arrived-eat-225931289.html", "https://developer.ideogram.ai/v1/api-reference/legacy-endpoints/generate", "https://artificialanalysis.ai/image/leaderboard/text-to-image" ], "unverified": [] }, { "id": "idsv-ft-whisper", "name": "FT-whisper (IDSV)", "level": "model", "family_id": null, "developer": [ "institut-teknologi-bandung", "universitas-indonesia", "jaist", "naist" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "autoregressive", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-10-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Indonesian", "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Whisper fine-tuned on the Indonesian IDSV dataset in the ITB/UI/JAIST/NAIST study.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2410.08828v2" ], "unverified": [ "indonesia_hosting" ], "indonesian_measured": [ { "benchmark": "IDSV Indonesian speech-variability test set, all groups (fine-tuned whisper-small)", "score": "14.85", "metric": "WER %, lower is better", "run_by": "paper authors (arXiv 2410.08828)", "url": "https://arxiv.org/html/2410.08828v2" } ] }, { "id": "idsv-mms-ft", "name": "MMS fine-tuned, no LM (IDSV)", "level": "model", "family_id": null, "developer": [ "institut-teknologi-bandung", "universitas-indonesia", "jaist", "naist" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-10-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Indonesian", "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MMS fine-tuned on IDSV without a language model.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2410.08828v2" ], "unverified": [ "indonesia_hosting" ], "indonesian_measured": [ { "benchmark": "IDSV Indonesian speech-variability test set, all groups (fine-tuned mms-300m, no language model)", "score": "29.66", "metric": "WER %, lower is better", "run_by": "paper authors (arXiv 2410.08828)", "url": "https://arxiv.org/html/2410.08828v2" } ] }, { "id": "idsv-mms-ft-lm", "name": "MMS fine-tuned with LM (IDSV)", "level": "model", "family_id": null, "developer": [ "institut-teknologi-bandung", "universitas-indonesia", "jaist", "naist" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-10-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Indonesian", "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MMS fine-tuned on IDSV with a language model.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2410.08828v2" ], "unverified": [ "indonesia_hosting" ], "indonesian_measured": [ { "benchmark": "IDSV Indonesian speech-variability test set, all groups (fine-tuned mms-300m with language model)", "score": "19.95", "metric": "WER %, lower is better", "run_by": "paper authors (arXiv 2410.08828)", "url": "https://arxiv.org/html/2410.08828v2" } ] }, { "id": "iflytek-spark", "name": "iFlytek Spark", "level": "model", "family_id": null, "developer": [ "iflytek" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Not found on AA or Arena pages read.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/leaderboards/models" ], "unverified": [ "status" ] }, { "id": "inception-jais-2", "name": "Jais 2", "level": "model", "family_id": null, "developer": [ "inception", "mbzuai", "g42" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-08", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Arabic", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Arabic LLM from Inception, MBZUAI and G42 (paper about Aug 2026); the inceptionai HF org shows 0 models.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/inceptionai" ], "unverified": [ "licence", "openness" ] }, { "id": "inception-labs-mercury", "name": "Mercury", "level": "model", "family_id": "inception-labs-mercury-family", "developer": [ "inception-labs" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "diffusion", "architecture_detail": "Transformer-parameterised diffusion LLM trained to predict multiple tokens in parallel; first release was Mercury Coder (Mini and Small)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-02", "release_date_basis": "inferred from page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "Paper reports 1109 tokens/s (Coder Mini) and 737 tokens/s (Coder Small) on H100, measured by Artificial Analysis", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "First commercially available diffusion-based LLM from Inception, launched in February 2025 with the Mercury Coder models.", "verified_on": "2026-10-11", "sources": [ "https://inceptionlabs.ai/blog/mercury-refreshed", "https://arxiv.org/abs/2506.17298" ], "unverified": [] }, { "id": "inception-labs-mercury-2", "name": "Mercury 2", "level": "model", "family_id": "inception-labs-mercury-family", "developer": [ "inception-labs" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "diffusion", "architecture_detail": "Diffusion language model that generates in parallel rather than decoding token by token; tunable reasoning", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 128000, "release_date": "2026-03-04", "release_date_basis": "registry date", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "consumer app" ], "deployment_tiers": [], "footprint_note": "Vendor reports 1,009 tokens per second on NVIDIA Blackwell GPUs; OpenAI API compatible", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Inception reasoning language model built on diffusion for real-time latency, with tool use and schema-aligned JSON output.", "verified_on": "2026-10-11", "sources": [ "https://www.inceptionlabs.ai/blog/introducing-mercury-2", "https://openrouter.ai/inception/mercury-2" ], "unverified": [] }, { "id": "inception-labs-mercury-2-5", "name": "Mercury 2.5", "level": "model", "family_id": "inception-labs-mercury-family", "developer": [ "inception-labs" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "diffusion", "architecture_detail": "Diffusion language model (dLLM) that produces and refines multiple tokens in parallel", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 260000, "release_date": "2026-09-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "Vendor reports 1,107 tokens per second on widely available NVIDIA GPUs", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Inception diffusion LLM with tunable reasoning, served through the Inception API, Baseten and OpenRouter; vendor reports a 40% intelligence gain over Mercury 2.", "verified_on": "2026-10-11", "sources": [ "https://www.inceptionlabs.ai/blog/introducing-mercury-2-5", "https://www.inceptionlabs.ai/blog", "https://openrouter.ai/api/v1/models" ], "unverified": [] }, { "id": "inception-labs-mercury-family", "name": "Mercury family", "level": "family", "family_id": null, "developer": [ "inception-labs" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Mercury is the Inception family of diffusion large language models (dLLMs) that generate tokens in parallel.", "verified_on": "2026-10-11", "sources": [ "https://www.inceptionlabs.ai/blog", "https://arxiv.org/abs/2506.17298", "https://www.inceptionlabs.ai/blog/introducing-mercury-2" ], "unverified": [] }, { "id": "inclusionai-ling-1t", "name": "Ling-1T", "level": "model", "family_id": "inclusionai-ling-2-0", "developer": [ "inclusionai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "MoE with 1/32 activation ratio, MTP layers, aux-loss-free sigmoid routing, QK Normalization, GQA; FP8 mixed-precision training", "params_total": 1000000000000, "params_active": 50000000000, "params_checkpoint": null, "params_note": "About 50B active per token", "context_length": 131072, "release_date": "2025-10-09", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "inclusionAI trillion-parameter non-thinking MoE model, open-sourced by Ant Group in October 2025.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/inclusionAI/Ling-1T", "https://www.artificialintelligence-news.com/news/trillion-parameter-ai-model-ant-group-ling-1t/" ], "unverified": [] }, { "id": "inclusionai-ling-2-0", "name": "Ling 2.0", "level": "family", "family_id": null, "developer": [ "inclusionai" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Ling 2.0 is the inclusionAI series of sparse MoE language models, with Ling non-thinking and Ring thinking models.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/inclusionAI/Ring-flash-2.0", "https://huggingface.co/inclusionAI/Ling-flash-2.0", "https://huggingface.co/inclusionAI/Ling-1T" ], "unverified": [] }, { "id": "inclusionai-ling-3", "name": "Ling 3", "level": "family", "family_id": null, "developer": [ "inclusionai" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Ling 3 is the inclusionAI series of hybrid-linear MoE language models, including domain and vision variants.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/inclusionAI/Ling-3.0-flash", "https://openrouter.ai/api/v1/models" ], "unverified": [] }, { "id": "inclusionai-ling-3-0-flash", "name": "Ling 3.0 Flash (Ling-3.0-flash)", "level": "model", "family_id": "inclusionai-ling-3", "developer": [ "inclusionai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Hybrid-linear MoE: 35 KDA layers and 7 Gated MLA layers (5:1), 512 routed experts, 1 shared, 8 active", "params_total": 124000000000, "params_active": 5100000000, "params_checkpoint": null, "params_note": "Safetensors metadata lists 127B", "context_length": 262144, "release_date": "2026-07-23", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "API", "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "inclusionAI 124B-total, 5.1B-active MoE model designed for token efficiency and production-scale agentic inference.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/inclusionAI/Ling-3.0-flash", "https://openrouter.ai/api/v1/models" ], "unverified": [] }, { "id": "inclusionai-ling-3-0-flash-fin", "name": "Ling 3.0 Flash Fin (Ling-3.0-flash-Fin)", "level": "model", "family_id": "inclusionai-ling-3", "developer": [ "inclusionai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": null, "params_total": 124000000000, "params_active": 5100000000, "params_checkpoint": null, "params_note": null, "context_length": 262144, "release_date": "2026-08-27", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "API", "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Finance-enhanced model built on Ling 3.0 Flash for long-horizon financial research, search, valuation and spreadsheet agent workflows.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/inclusionAI/Ling-3.0-flash-FIN", "https://openrouter.ai/api/v1/models" ], "unverified": [] }, { "id": "inclusionai-ling-3-0-flash-vl", "name": "Ling 3.0 Flash VL (Ling-3.0-flash-VL)", "level": "model", "family_id": "inclusionai-ling-3", "developer": [ "inclusionai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "42-layer hybrid KDA and Gated MLA MoE backbone (5:1) with ViT encoder, two-layer MLP projector and VideoRoPE", "params_total": 124000000000, "params_active": 5500000000, "params_checkpoint": null, "params_note": null, "context_length": 262144, "release_date": "2026-09-10", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "API", "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vision-language model built on Ling 3.0 Flash that adds native image and video perception.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/inclusionAI/Ling-3.0-flash-VL", "https://openrouter.ai/api/v1/models" ], "unverified": [] }, { "id": "inclusionai-ling-3-1-flash", "name": "Ling 3.1 Flash", "level": "model", "family_id": "inclusionai-ling-3", "developer": [ "inclusionai" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "Hybrid reasoning mixture-of-experts", "params_total": 560000000000, "params_active": 25000000000, "params_checkpoint": null, "params_note": "From the OpenRouter listing; no developer model card found", "context_length": 262144, "release_date": "2026-10-02", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "inclusionAI hybrid reasoning MoE model with 25B active of 560B total parameters, served through OpenRouter.", "verified_on": "2026-10-11", "sources": [ "https://openrouter.ai/api/v1/models", "https://openrouter.ai/inclusionai/ling-3.1-flash" ], "unverified": [ "params_total", "params_active" ] }, { "id": "inclusionai-ling-flash-2-0", "name": "Ling-flash-2.0", "level": "model", "family_id": "inclusionai-ling-2-0", "developer": [ "inclusionai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "MoE, 1/32 activation ratio, aux-loss-free sigmoid routing, MTP layers, QK-Norm, Partial-RoPE", "params_total": 100000000000, "params_active": 6100000000, "params_checkpoint": null, "params_note": "4.8B non-embedding active; safetensors lists 103B", "context_length": 32768, "release_date": "2025-09-17", "release_date_basis": "registry date", "weights_date": "2025-09-17", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": "Card: 32K context, extendable to 128K with YaRN; 200+ tokens/s on H20", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "inclusionAI 100B-total, 6.1B-active MoE language model trained on 20T+ tokens with SFT and multi-stage RL.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/inclusionAI/Ling-flash-2.0", "https://huggingface.co/api/models/inclusionAI/Ling-flash-2.0" ], "unverified": [] }, { "id": "inclusionai-ling-flash-base-2-0", "name": "Ling-flash-base-2.0", "level": "model", "family_id": "inclusionai-ling-2-0", "developer": [ "inclusionai" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Base model named as the base of Ling-flash-2.0 and Ring-flash-2.0.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/inclusionAI/Ling-flash-2.0", "https://huggingface.co/inclusionAI/Ring-flash-2.0" ], "unverified": [] }, { "id": "inclusionai-ring-flash-2-0", "name": "Ring-flash-2.0", "level": "model", "family_id": "inclusionai-ling-2-0", "developer": [ "inclusionai" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "MoE, 1/32 activation ratio, MTP layers", "params_total": 100000000000, "params_active": 6100000000, "params_checkpoint": null, "params_note": "Safetensors lists 103B", "context_length": null, "release_date": "2025-09-19", "release_date_basis": "registry date", "weights_date": "2025-09-19", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "inclusionAI thinking model optimised from Ling-flash-2.0-base for math, code and reasoning.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/inclusionAI/Ring-flash-2.0", "https://huggingface.co/api/models/inclusionAI/Ring-flash-2.0" ], "unverified": [] }, { "id": "indextts-2-5", "name": "IndexTTS-2.5", "level": "model", "family_id": null, "developer": [], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open TTS model trending on Hugging Face (2025 to 2026).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/models?pipeline_tag=text-to-speech&sort=trending" ], "unverified": [] }, { "id": "indonesian-nlp-wav2vec2-indonesian-javanese-sundanese", "name": "wav2vec2-indonesian-javanese-sundanese", "level": "model", "family_id": null, "developer": [ "indonesian-nlp" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "wav2vec2-large-xlsr-53 fine-tuned with CTC", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-03-02", "release_date_basis": "registry date", "weights_date": "2022-03-02", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Indonesian (Common Voice), Javanese (OpenSLR SLR41) and Sundanese (OpenSLR SLR44)", "indonesian_support": "measured on a benchmark", "regional_languages": [ "Javanese", "Sundanese" ], "indonesia_hosting": "unknown", "status": null, "summary": "A speech recognition model for Indonesian, Javanese and Sundanese from the Multilingual Speech Recognition for Indonesian Languages project, fine-tuned from wav2vec2-large-xlsr-53.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/indonesian-nlp/wav2vec2-indonesian-javanese-sundanese", "https://huggingface.co/api/models/indonesian-nlp/wav2vec2-indonesian-javanese-sundanese" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "indonlp-cendol", "name": "Cendol", "level": "family", "family_id": null, "developer": [ "indonlp" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 13000000000, "params_active": null, "params_checkpoint": null, "params_note": "0.3B to 13B", "context_length": null, "release_date": "2024-04-21", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "consumer GPU", "CPU server" ], "footprint_note": null, "runtimes": [], "languages_note": "Indonesian + local languages", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "IndoNLP's instruction-tuned mT5 (small to XXL) and Llama 2 (7B, 13B) models for Indonesian and local languages.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/indonlp" ], "unverified": [ "licence", "indonesian_measured", "indonesia_hosting" ] }, { "id": "indonlp-indobert", "name": "IndoBERT", "level": "family", "family_id": null, "developer": [ "indonlp" ], "model_class": "embedding", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Indonesian BERT encoders from IndoNLP (ITB, HKUST, Gojek researchers), evaluated on IndoNLU (2020); widely used academic baseline.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/indobenchmark/indobert-base-p1", "https://arxiv.org/pdf/2009.05387" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "indonlp-indobert-base", "name": "IndoBERT base", "level": "model", "family_id": "indonlp-indobert", "developer": [ "indonlp" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": "BERT trained on Indo4B (23.43 GB)", "params_total": 124500000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "IndoBERT base encoder for Indonesian classification and NER.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/indobenchmark/indobert-base-p1", "https://arxiv.org/pdf/2009.05387" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "indonlp-indobert-large", "name": "IndoBERT large", "level": "model", "family_id": "indonlp-indobert", "developer": [ "indonlp" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": "BERT trained on Indo4B (23.43 GB)", "params_total": 335200000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "IndoBERT large encoder.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2009.05387" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "indonlp-indobert-lite", "name": "IndoBERT lite", "level": "model", "family_id": "indonlp-indobert", "developer": [ "indonlp" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": "BERT trained on Indo4B (23.43 GB)", "params_total": 11700000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "IndoBERT lite encoder.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2009.05387" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "inria-tabicl", "name": "TabICL", "level": "model", "family_id": null, "developer": [ "inria" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "TabICL v1, scaled to about 500k samples.", "verified_on": "2026-10-10", "sources": [ "https://mindfulmodeler.substack.com/p/the-state-of-tabular-foundation-models" ], "unverified": [] }, { "id": "inria-tabicl-v2", "name": "TabICL v2", "level": "model", "family_id": null, "developer": [ "inria" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "dense transformer", "architecture_detail": "In-context transformer", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026", "release_date_basis": "notes", "weights_date": null, "openness": "open source", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": "Median train about 4.0 s", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "INRIA soda team tabular in-context learner; rank 4 on TabArena.", "verified_on": "2026-10-10", "sources": [ "https://benchmarklist.com/benchmarks/tabarena_all/", "https://mindfulmodeler.substack.com/p/the-state-of-tabular-foundation-models" ], "unverified": [] }, { "id": "institute-of-foundation-models-k2-horizon-375b-a23b", "name": "K2 Horizon 375B A23B", "level": "model", "family_id": null, "developer": [ "institute-of-foundation-models" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 375000000000, "params_active": 23000000000, "params_checkpoint": null, "params_note": "From model name", "context_length": 524000, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/leaderboards/models" ], "unverified": [ "developer" ] }, { "id": "institute-of-science-tokyo-gpt-oss-swallow", "name": "GPT-OSS-Swallow", "level": "model", "family_id": null, "developer": [ "institute-of-science-tokyo" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Swallow models built on gpt-oss bases; licence inherited from base.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/tokyotech-llm", "https://huggingface.co/openai/gpt-oss-120b" ], "unverified": [] }, { "id": "institute-of-science-tokyo-gpt-oss-swallow-20b", "name": "GPT-OSS-Swallow-20B", "level": "model", "family_id": null, "developer": [ "institute-of-science-tokyo" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/openai", "https://huggingface.co/openai/gpt-oss-safeguard-20b" ], "unverified": [] }, { "id": "institute-of-science-tokyo-medical-swallow", "name": "Medical-Swallow", "level": "model", "family_id": null, "developer": [ "institute-of-science-tokyo" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/tokyotech-llm" ], "unverified": [] }, { "id": "institute-of-science-tokyo-qwen3-swallow", "name": "Qwen3-Swallow", "level": "model", "family_id": null, "developer": [ "institute-of-science-tokyo" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Swallow line spans 8B to 120B", "context_length": null, "release_date": "2026", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Swallow models built on Qwen3 bases; licence inherited from base.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/tokyotech-llm" ], "unverified": [] }, { "id": "intellindust-deim", "name": "DEIM", "level": "model", "family_id": null, "developer": [ "intellindust-ai-lab", "city-university-of-hong-kong", "great-bay-university", "hefei-normal-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": null, "architecture": "dense transformer", "architecture_detail": "DETR with Improved Matching for Fast Convergence; a training scheme applied on top of D-FINE", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-12-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time DETR training scheme; the paper compares DEIM-D-FINE-L against D-FINE-L.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2412.04234", "https://github.com/ShihuaHuang95/DEIM", "https://playground.roboflow.com/models/compare/deim-vs-rf-detr" ], "unverified": [] }, { "id": "internscience-agents-a1-4b", "name": "Agents-A1-4B", "level": "model", "family_id": null, "developer": [ "internscience" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "4B (per name)", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "InternScience 4B agent model with the only GAIA row on BenchLM (self-report).", "verified_on": "2026-10-10", "sources": [ "https://benchlm.ai/benchmarks/gaia" ], "unverified": [] }, { "id": "ipoc", "name": "IPOC", "level": "model", "family_id": null, "developer": [], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Video model listed on the VBench leaderboard; developer not stated in the notes.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/spaces/Vchitect/VBench_Leaderboard" ], "unverified": [] }, { "id": "ipow", "name": "IPOW", "level": "model", "family_id": null, "developer": [], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Video model listed on the VBench leaderboard; developer not stated in the notes.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/spaces/Vchitect/VBench_Leaderboard" ], "unverified": [] }, { "id": "isolation-forest", "name": "Isolation forest", "level": "model", "family_id": null, "developer": [], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Anomaly detection baseline on SKAB.", "verified_on": "2026-10-10", "sources": [ "https://github.com/waico/SKAB" ], "unverified": [] }, { "id": "jetbrains-mellum2-1-12b-a2-5b-thinking", "name": "Mellum2.1-12B-A2.5B-Thinking", "level": "model", "family_id": null, "developer": [ "jetbrains" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 12000000000, "params_active": 2500000000, "params_checkpoint": null, "params_note": null, "context_length": 131000, "release_date": "2026-10-08", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "laptop" ], "footprint_note": "T5 estimate (unverified). GGUF available. Tech report May 29.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "JetBrains Mellum2 line small MoE coding model with thinking mode.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/JetBrains", "https://huggingface.co/JetBrains/Mellum2.1-12B-A2.5B-Thinking" ], "unverified": [ "deployment_tiers", "release_date" ] }, { "id": "jina-ai-jina-embeddings-v3", "name": "jina-embeddings-v3", "level": "model", "family_id": null, "developer": [ "jina-ai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [], "architecture": "other", "architecture_detail": "Jina-XLM-RoBERTa encoder with RoPE and 5 task LoRA adapters; Matryoshka dimensions 32 to 1024", "params_total": 572310396, "params_active": null, "params_checkpoint": null, "params_note": "Hugging Face API Safetensors count", "context_length": 8192, "release_date": "2024-09-05", "release_date_basis": "registry date", "weights_date": "2024-09-05", "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "API", "cloud marketplace", "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Foundation supports 100 languages; tuning focused on 30 including Indonesian, Thai and Vietnamese", "indonesian_support": "listed by developer", "regional_languages": [ "Javanese", "Sundanese" ], "indonesia_hosting": "unknown", "status": null, "summary": "jina-embeddings-v3 is a multilingual multi-task text embedding model with task-specific LoRA adapters and 8192-token input.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/jinaai/jina-embeddings-v3", "https://huggingface.co/api/models/jinaai/jina-embeddings-v3" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "jina-ai-jina-embeddings-v5-omni-nano", "name": "jina-embeddings-v5-omni-nano", "level": "model", "family_id": null, "developer": [ "jina-ai" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 900000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-05-07", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "119+ languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Jina AI jina-embeddings-v5-omni-nano; CC BY-NC 4.0 (commercial use needs a sales licence).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/jinaai" ], "unverified": [ "modalities_in" ] }, { "id": "jina-ai-jina-embeddings-v5-omni-small", "name": "jina-embeddings-v5-omni-small", "level": "model", "family_id": null, "developer": [ "jina-ai" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 2000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-05-07", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "119+ languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Jina AI jina-embeddings-v5-omni-small; CC BY-NC 4.0 (commercial use needs a sales licence).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/jinaai" ], "unverified": [ "modalities_in" ] }, { "id": "jina-ai-jina-embeddings-v5-text-nano", "name": "jina-embeddings-v5-text-nano", "level": "model", "family_id": null, "developer": [ "jina-ai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 32000, "release_date": "2026-02-18", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "119+ languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Jina AI jina-embeddings-v5-text-nano; CC BY-NC 4.0 (commercial use needs a sales licence). 1024-dimension Matryoshka.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/jinaai" ], "unverified": [] }, { "id": "jina-ai-jina-embeddings-v5-text-small", "name": "jina-embeddings-v5-text-small", "level": "model", "family_id": null, "developer": [ "jina-ai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 677000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 32000, "release_date": "2026-02-18", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "119+ languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Jina AI jina-embeddings-v5-text-small; CC BY-NC 4.0 (commercial use needs a sales licence). 1024-dimension Matryoshka.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/jinaai", "https://huggingface.co/jinaai/jina-embeddings-v5-text-small" ], "unverified": [] }, { "id": "jina-ai-jina-reranker-v3-5", "name": "jina-reranker-v3.5", "level": "model", "family_id": null, "developer": [ "jina-ai" ], "model_class": "reranker", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 600000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-08-03", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "119+ languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Jina AI jina-reranker-v3.5; CC BY-NC 4.0 (commercial use needs a sales licence).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/jinaai" ], "unverified": [] }, { "id": "jt-cv", "name": "JT-CV", "level": "model", "family_id": null, "developer": [], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Video model listed on the VBench leaderboard; developer not stated in the notes.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/spaces/Vchitect/VBench_Leaderboard" ], "unverified": [] }, { "id": "jt3-5", "name": "JT3.5", "level": "model", "family_id": null, "developer": [], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Video model listed on the VBench leaderboard; developer not stated in the notes.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/spaces/Vchitect/VBench_Leaderboard" ], "unverified": [] }, { "id": "k2-fsa-omnivoice", "name": "OmniVoice", "level": "model", "family_id": null, "developer": [ "k2-fsa" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "diffusion", "architecture_detail": "Diffusion-LM style, base Qwen3-0.6B", "params_total": 600000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-unspecified", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": "600+ languages claimed; Indonesian not specifically confirmed", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "k2-fsa open TTS model; code Apache 2.0, weights CC-BY-NC.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/k2-fsa/OmniVoice" ], "unverified": [] }, { "id": "keisler-2022", "name": "Keisler (2022)", "level": "model", "family_id": null, "developer": [], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-02-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Weather model on WeatherBench 2 (2020).", "verified_on": "2026-10-10", "sources": [ "https://researchsites.withgoogle.com/weatherbench/deterministic/", "https://arxiv.org/abs/2202.07575" ], "unverified": [] }, { "id": "khan-et-al-extra-trees-oil-palm-yield", "name": "Extra Trees oil palm yield model (JOPR 2025)", "level": "model", "family_id": null, "developer": [], "model_class": "tabular", "modalities_in": [], "modalities_out": [], "architecture": "other", "architecture_detail": "Extra Trees regression; compared with Random Forest, Gradient Boosting and Linear Regression", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "research only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Academic oil palm yield model on Pahang agromet data 1986 to 2020; R2 0.88 (ET) vs 0.79 RF, 0.78 GB, 0.41 LR (Khan, Kamaruddin, Sheikh, Rahman, Bakht, Malaysia).", "verified_on": "2026-10-10", "sources": [ "https://jopr.mpob.gov.my/prediction-of-oil-palm-yield-using-machine-learning-comparison-of-linear-and-non-linear-algorithms-with-multivariate-time-series-data/" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "krea-2", "name": "Krea 2", "level": "family", "family_id": null, "developer": [ "krea" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Krea 2 image models (Raw, Turbo, Large).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/krea/Krea-2-Turbo" ], "unverified": [] }, { "id": "krea-2-large", "name": "Krea 2 Large", "level": "model", "family_id": "krea-2", "developer": [ "krea" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-05", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Krea 2 Large, available via Artificial Analysis arena.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text", "https://huggingface.co/krea/Krea-2-Turbo" ], "unverified": [] }, { "id": "krea-2-raw", "name": "Krea 2 Raw", "level": "model", "family_id": "krea-2", "developer": [ "krea" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": "DiT", "params_total": 12000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-06-22", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "krea-2-community", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Krea 12B DiT image model, gated open weights under the Krea 2 Community Licence.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/krea/Krea-2-Turbo", "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text" ], "unverified": [ "deployment_tiers" ] }, { "id": "krea-2-turbo", "name": "Krea 2 Turbo", "level": "model", "family_id": "krea-2", "developer": [ "krea" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": "DiT, distilled from Raw", "params_total": 12000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-06-22", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "krea-2-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Distilled Krea 2 image model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/krea/Krea-2-Turbo" ], "unverified": [] }, { "id": "krea-flux-1-krea-dev", "name": "FLUX.1 Krea Dev", "level": "model", "family_id": null, "developer": [ "krea", "black-forest-labs" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-07", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier Krea image model built with BFL.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Black_Forest_Labs", "https://huggingface.co/krea/Krea-2-Turbo" ], "unverified": [] }, { "id": "kuaishou-kling", "name": "Kling", "level": "family", "family_id": null, "developer": [ "kuaishou" ], "model_class": "video generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Kuaishou Kling video model line.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Kling_AI" ], "unverified": [] }, { "id": "kuaishou-kling-1-6", "name": "Kling 1.6", "level": "model", "family_id": "kuaishou-kling", "developer": [ "kuaishou" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier Kling model.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Kling_AI", "https://huggingface.co/spaces/Vchitect/VBench_Leaderboard" ], "unverified": [] }, { "id": "kuaishou-kling-2-0", "name": "Kling 2.0", "level": "model", "family_id": "kuaishou-kling", "developer": [ "kuaishou" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier Kling model.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Kling_AI" ], "unverified": [] }, { "id": "kuaishou-kling-2-1", "name": "Kling 2.1", "level": "model", "family_id": "kuaishou-kling", "developer": [ "kuaishou" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-05", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier Kling model.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Kling_AI" ], "unverified": [] }, { "id": "kuaishou-kling-3-0", "name": "Kling 3.0", "level": "model", "family_id": "kuaishou-kling", "developer": [ "kuaishou" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": "DiT plus 3D VAE, full spatiotemporal attention", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-02-07", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Kuaishou video model; the 1080p Pro version is the Artificial Analysis anchor at 1000. Kling outputs carry an invisible watermark aligned with open provenance standards.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Kling_AI", "https://artificialanalysis.ai/video/leaderboard/text-to-video", "https://kling.ai/docs/ai-content-detection" ], "unverified": [] }, { "id": "kuaishou-kling-3-0-omni", "name": "Kling 3.0 Omni", "level": "variant", "family_id": "kuaishou-kling-3-0", "developer": [ "kuaishou" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-02", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Omni version of Kling 3.0.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/video/leaderboard/text-to-video", "https://en.wikipedia.org/wiki/Kling_AI" ], "unverified": [] }, { "id": "kuaishou-kolors", "name": "Kolors", "level": "model", "family_id": null, "developer": [ "kuaishou" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": "Latent diffusion, SDXL pipeline, ChatGLM3 text encoder", "params_total": 3000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Kuaishou 3B latent diffusion image model; code Apache 2.0, commercial use by registration.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Kwai-Kolors/Kolors", "https://en.wikipedia.org/wiki/Kuaishou" ], "unverified": [ "deployment_tiers" ] }, { "id": "kuaishou-onerec", "name": "OneRec (V1)", "level": "model", "family_id": null, "developer": [ "kuaishou" ], "model_class": "recommendation", "modalities_in": [ "action" ], "modalities_out": [ "action" ], "architecture": "MoE", "architecture_detail": "End-to-end generative encoder-decoder with sparse MoE; session-wise generation; Iterative Preference Alignment with DPO and a reward model", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-02-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "internal only" ], "deployment_tiers": [ "hyperscale cluster" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Kuaishou generative recommender deployed in the main scene, +1.6% watch time.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2502.18965", "https://arxiv.org/pdf/2502.18965" ], "unverified": [] }, { "id": "kuaishou-onerec-foundation-1-7b", "name": "OneRec-Foundation 1.7B", "level": "variant", "family_id": "kuaishou-openonerec", "developer": [ "kuaishou" ], "model_class": "recommendation", "modalities_in": [ "action" ], "modalities_out": [ "action" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 1700000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OneRec-Foundation size variant (Standard and Pro).", "verified_on": "2026-10-10", "sources": [ "https://github.com/Kuaishou-OneRec/OpenOneRec" ], "unverified": [] }, { "id": "kuaishou-onerec-foundation-8b", "name": "OneRec-Foundation 8B", "level": "variant", "family_id": "kuaishou-openonerec", "developer": [ "kuaishou" ], "model_class": "recommendation", "modalities_in": [ "action" ], "modalities_out": [ "action" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OneRec-Foundation size variant (Standard and Pro).", "verified_on": "2026-10-10", "sources": [ "https://github.com/Kuaishou-OneRec/OpenOneRec" ], "unverified": [] }, { "id": "kuaishou-onerec-think", "name": "OneRec-Think", "level": "model", "family_id": null, "developer": [ "kuaishou" ], "model_class": "recommendation", "modalities_in": [ "action" ], "modalities_out": [ "action" ], "architecture": "other", "architecture_detail": "Integrates dialogue, reasoning and recommendation: itemic alignment, reasoning scaffolding, recommendation-specific reward", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Reasoning-integrated OneRec; results not read.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2510.11639" ], "unverified": [ "developer" ] }, { "id": "kuaishou-onerec-v2", "name": "OneRec-V2", "level": "model", "family_id": null, "developer": [ "kuaishou" ], "model_class": "recommendation", "modalities_in": [ "action" ], "modalities_out": [ "action" ], "architecture": "dense transformer", "architecture_detail": "Lazy Decoder-Only Architecture removing the encoder; Duration-Aware Reward Shaping, Adaptive Ratio Clipping", "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": "Scaled to 8B", "context_length": null, "release_date": "2025-08-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "internal only" ], "deployment_tiers": [ "hyperscale cluster" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Cuts total compute 94% and training resources 90% versus V1; Kuaishou A/B App Stay Time +0.467% / +0.741%.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2508.20900" ], "unverified": [] }, { "id": "kuaishou-openonerec", "name": "OpenOneRec / OneRec-Foundation", "level": "family", "family_id": null, "developer": [ "kuaishou" ], "model_class": "recommendation", "modalities_in": [ "action" ], "modalities_out": [ "action" ], "architecture": "dense transformer", "architecture_detail": "Built on Qwen3-1.7B and Qwen3-8B backbones; Standard trained on about 33B tokens of open data, Pro on about 130B tokens including Kuaishou industrial corpus", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "LLM-backbone recommender family (1.7B, 8B; Standard and Pro) shipping RecIF-Bench.", "verified_on": "2026-10-10", "sources": [ "https://github.com/Kuaishou-OneRec/OpenOneRec", "https://huggingface.co/api/models/OpenOneRec/OneRec-8B" ], "unverified": [] }, { "id": "kumo-ai-kumorfm", "name": "KumoRFM", "level": "model", "family_id": null, "developer": [ "kumo-ai" ], "model_class": "graph", "modalities_in": [ "table", "graph" ], "modalities_out": null, "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Relational foundation model from Kumo AI; no details fetched.", "verified_on": "2026-10-10", "sources": [ "https://mindfulmodeler.substack.com/p/the-state-of-tabular-foundation-models" ], "unverified": [] }, { "id": "kwaikat-kat-coder", "name": "KAT-Coder", "level": "family", "family_id": null, "developer": [ "kwaikat" ], "model_class": "code", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "KAT-Coder is the KwaiKAT series of agentic coding models.", "verified_on": "2026-10-11", "sources": [ "https://artificialanalysis.ai/models/kat-coder-pro-v1" ], "unverified": [] }, { "id": "kwaikat-kat-coder-pro-v1", "name": "KAT-Coder-Pro V1", "level": "model", "family_id": "kwaikat-kat-coder", "developer": [ "kwaikat" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 256000, "release_date": "2025-11-11", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "KwaiKAT agentic coding model; Artificial Analysis lists it as deprecated, superseded by KAT Coder Pro V2.", "verified_on": "2026-10-11", "sources": [ "https://artificialanalysis.ai/models/kat-coder-pro-v1" ], "unverified": [] }, { "id": "layer-6-ai-tabdpt", "name": "TabDPT", "level": "model", "family_id": null, "developer": [ "layer-6-ai" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": "ICL + self-supervised", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024", "release_date_basis": "notes", "weights_date": "2025-06-18", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": "Median predict about 287 s", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Layer 6 AI (TD Bank) tabular foundation model.", "verified_on": "2026-10-10", "sources": [ "https://benchmarklist.com/benchmarks/tabarena_all/", "https://mindfulmodeler.substack.com/p/the-state-of-tabular-foundation-models", "https://huggingface.co/api/models/Layer6/TabDPT" ], "unverified": [] }, { "id": "lexsi-labs-orion-bix", "name": "Orion-Bix", "level": "model", "family_id": null, "developer": [ "lexsi-labs" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": "Bi-axial attention", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Lexsi Labs tabular model.", "verified_on": "2026-10-10", "sources": [ "https://mindfulmodeler.substack.com/p/the-state-of-tabular-foundation-models" ], "unverified": [ "licence" ] }, { "id": "lexsi-labs-orion-msp", "name": "Orion-MSP", "level": "model", "family_id": null, "developer": [ "lexsi-labs" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": "Sparse multi-scale attention", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025", "release_date_basis": "notes", "weights_date": "2025-11-03", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Lexsi Labs tabular model.", "verified_on": "2026-10-10", "sources": [ "https://mindfulmodeler.substack.com/p/the-state-of-tabular-foundation-models", "https://huggingface.co/api/models/Lexsi/Orion-MSP" ], "unverified": [] }, { "id": "lg-ai-research-exaone", "name": "EXAONE", "level": "family", "family_id": null, "developer": [ "lg-ai-research" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "LG AI Research model family with Naphtha, Business Intelligence and Chemical Agent applications; licence unverified.", "verified_on": "2026-10-10", "sources": [ "https://www.lgresearch.ai/exaone" ], "unverified": [ "licence", "indonesian_measured", "indonesia_hosting" ] }, { "id": "lg-ai-research-exaone-4-0-32b", "name": "EXAONE 4.0 (32B)", "level": "model", "family_id": "lg-ai-research-exaone", "developer": [ "lg-ai-research" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "64 layers, GQA (40 query, 8 KV heads), hybrid local sliding-window and global attention at 3:1, QK-Reorder-Norm; unified reasoning and non-reasoning modes", "params_total": 32000000000, "params_active": null, "params_checkpoint": null, "params_note": "30.95B without embeddings", "context_length": 131072, "release_date": "2025-07-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "exaone-ai-model-license-1-2-nc", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English, Korean, Spanish", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "LG AI Research 32B model that unifies reasoning and non-reasoning modes; weights are under a non-commercial research licence.", "verified_on": "2026-10-11", "sources": [ "https://arxiv.org/abs/2507.11407", "https://huggingface.co/LGAI-EXAONE/EXAONE-4.0-32B", "https://huggingface.co/LGAI-EXAONE/EXAONE-4.0-32B/blob/main/LICENSE" ], "unverified": [] }, { "id": "lg-ai-research-exaone-4-5", "name": "EXAONE 4.5", "level": "model", "family_id": "lg-ai-research-exaone", "developer": [ "lg-ai-research" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": 33000000000, "params_active": null, "params_checkpoint": null, "params_note": "33B on Hugging Face", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "LG EXAONE model, 33B on Hugging Face; licence unverified.", "verified_on": "2026-10-10", "sources": [ "https://www.lgresearch.ai/exaone" ], "unverified": [ "licence" ] }, { "id": "lg-ai-research-exaone-4-5-33b", "name": "EXAONE 4.5-33B", "level": "model", "family_id": null, "developer": [ "lg-ai-research" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 33000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-07-06", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "consumer GPU" ], "footprint_note": "Consumer GPU at 4-bit (GGUF, AWQ)", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "LG's 33B dense model; licence not verified.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/LGAI-EXAONE" ], "unverified": [ "licence" ] }, { "id": "lg-ai-research-exaone-forecast", "name": "EXAONE-Forecast", "level": "model", "family_id": null, "developer": [ "lg-ai-research" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "LG AI Research zero-shot forecasting model on GIFT-Eval.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/spaces/Salesforce/GIFT-Eval" ], "unverified": [] }, { "id": "lg-ai-research-k-exaone", "name": "K-EXAONE", "level": "model", "family_id": null, "developer": [ "lg-ai-research" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 236000000000, "params_active": 23000000000, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/LGAI-EXAONE/K-EXAONE-2.0-750B-A37B" ], "unverified": [] }, { "id": "lg-ai-research-k-exaone-2-0", "name": "K-EXAONE 2.0", "level": "model", "family_id": null, "developer": [ "lg-ai-research" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "MoE (256 experts, 8 active + 1 shared), mostly SWA with periodic global layers, MTP", "params_total": 750000000000, "params_active": 37000000000, "params_checkpoint": null, "params_note": null, "context_length": 262144, "release_date": "2026-08-05", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "multi-GPU node" ], "footprint_note": "NVFP4 variant listed at 411B", "runtimes": [], "languages_note": "10 languages; Indonesian not among them", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "LG AI Research's 750B MoE text model, Apache 2.0 plus a stated ban on malicious use; reports a KGC-Safety score though it is not a guard model. Quants 3.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/leaderboards/models", "https://huggingface.co/LGAI-EXAONE/K-EXAONE-2.0-750B-A37B" ], "unverified": [ "release_date" ] }, { "id": "lightricks-ltx", "name": "LTX", "level": "family", "family_id": null, "developer": [ "lightricks" ], "model_class": "video generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Lightricks LTX video models (LTX-Video, LTXV, LTX-2, LTX-2.5).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Lightricks/LTX-2.5-Diffusers" ], "unverified": [] }, { "id": "lightricks-ltx-2", "name": "LTX-2", "level": "model", "family_id": "lightricks-ltx", "developer": [ "lightricks" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "LTX-2 (paper Jan 2026).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Lightricks/LTX-2.5-Diffusers" ], "unverified": [] }, { "id": "lightricks-ltx-2-5", "name": "LTX-2.5", "level": "model", "family_id": "lightricks-ltx", "developer": [ "lightricks" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video", "audio" ], "architecture": "diffusion", "architecture_detail": "DiT, joint audio-video, 2-stage with latent upsampler", "params_total": 19000000000, "params_active": null, "params_checkpoint": null, "params_note": "19B per HF card; 22B per Atlas Cloud", "context_length": null, "release_date": "2026-08-11", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "ltx-2-community", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": "Tens of GB; distilled download about 72 GB; offload suggested", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Lightricks open-weight audio-video model with 48 kHz audio, free under US$10M annual revenue.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Lightricks/LTX-2.5-Diffusers", "https://www.atlascloud.ai/de/blog/tips/is-wan-3.0-open-source", "https://artificialanalysis.ai/video/leaderboard/text-to-video", "https://en.wikipedia.org/wiki/Lightricks" ], "unverified": [] }, { "id": "lightricks-ltx-2-5-fast", "name": "LTX-2.5 Fast", "level": "variant", "family_id": "lightricks-ltx-2-5", "developer": [ "lightricks" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fast variant of LTX-2.5.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/video/leaderboard/text-to-video" ], "unverified": [] }, { "id": "lightricks-ltx-2-5-pro", "name": "LTX-2.5 Pro", "level": "variant", "family_id": "lightricks-ltx-2-5", "developer": [ "lightricks" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Pro variant of LTX-2.5.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/video/leaderboard/text-to-video" ], "unverified": [] }, { "id": "lightricks-ltx-video", "name": "LTX-Video", "level": "model", "family_id": "lightricks-ltx", "developer": [ "lightricks" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier Lightricks video model (also LTXV).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Lightricks/LTX-2.5-Diffusers" ], "unverified": [] }, { "id": "limo", "name": "LIMO", "level": "model", "family_id": null, "developer": [ "shanghai-jiao-tong-university" ], "model_class": "reasoning", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-02", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Reasoning fine-tune using about 1% of prior training data; developer not stated in the notes.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2502.03387", "https://arxiv.org/html/2502.03387" ], "unverified": [] }, { "id": "lingzhi-72b-chat", "name": "Lingzhi-72B-chat", "level": "model", "family_id": null, "developer": [], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "72B (per name)", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Model top on CMMLU five-shot.", "verified_on": "2026-10-10", "sources": [ "https://github.com/haonan-li/CMMLU" ], "unverified": [] }, { "id": "liquid-ai-d1-3b", "name": "d1-3B", "level": "model", "family_id": null, "developer": [ "liquid-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 3000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-10", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Liquid AI release, details unverified.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/LiquidAI" ], "unverified": [ "release_date", "summary" ] }, { "id": "liquid-ai-d1-omni-600m", "name": "d1-omni-600M", "level": "model", "family_id": null, "developer": [ "liquid-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 600000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-10", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Liquid AI release, details unverified.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/LiquidAI" ], "unverified": [ "release_date", "summary" ] }, { "id": "liquid-ai-lfm2-5", "name": "LFM2.5", "level": "family", "family_id": null, "developer": [ "liquid-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Double-gated convolution + GQA", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-08", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "lfm-1-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Liquid AI's LFM2.5 hybrid conv-attention models for CPU and edge; GGUF, ONNX and MLX variants.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/LiquidAI", "https://huggingface.co/LiquidAI/LFM2.5-8B-A1B" ], "unverified": [] }, { "id": "liquid-ai-lfm2-5-2-6b", "name": "LFM2.5-2.6B", "level": "model", "family_id": "liquid-ai-lfm2-5", "developer": [ "liquid-ai" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": null, "params_total": 2600000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 128000, "release_date": "2026-08", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "lfm-1-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "laptop", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/LiquidAI" ], "unverified": [] }, { "id": "liquid-ai-lfm2-5-230m", "name": "LFM2.5-230M", "level": "model", "family_id": "liquid-ai-lfm2-5", "developer": [ "liquid-ai" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": null, "params_total": 230000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 128000, "release_date": "2026-08", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "lfm-1-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "laptop", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/LiquidAI" ], "unverified": [] }, { "id": "liquid-ai-lfm2-5-350m", "name": "LFM2.5-350M", "level": "model", "family_id": "liquid-ai-lfm2-5", "developer": [ "liquid-ai" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": null, "params_total": 350000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 128000, "release_date": "2026-08", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "lfm-1-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "laptop", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/LiquidAI" ], "unverified": [] }, { "id": "liquid-ai-lfm2-5-8b-a1b", "name": "LFM2.5-8B-A1B", "level": "model", "family_id": "liquid-ai-lfm2-5", "developer": [ "liquid-ai" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "18 conv + 6 attention layers", "params_total": 8300000000, "params_active": 1500000000, "params_checkpoint": null, "params_note": null, "context_length": 128000, "release_date": "2026-08", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "lfm-1-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "laptop", "mobile or NPU" ], "footprint_note": "GGUF, ONNX and MLX variants", "runtimes": [], "languages_note": "10 languages listed; no Indonesian", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Liquid AI's small MoE hybrid model for edge use.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/LiquidAI", "https://huggingface.co/LiquidAI/LFM2.5-8B-A1B" ], "unverified": [] }, { "id": "liquid-ai-lfm2-5-encoder-230m", "name": "LFM2.5-Encoder 230M", "level": "model", "family_id": "liquid-ai-lfm2-5", "developer": [ "liquid-ai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 230000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-10", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "lfm-1-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Liquid AI encoder model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/LiquidAI" ], "unverified": [ "licence" ] }, { "id": "liquid-ai-lfm2-5-encoder-350m", "name": "LFM2.5-Encoder 350M", "level": "model", "family_id": "liquid-ai-lfm2-5", "developer": [ "liquid-ai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 350000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-10", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "lfm-1-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Liquid AI encoder model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/LiquidAI" ], "unverified": [ "licence" ] }, { "id": "liquid-ai-lfm2-5-thinking-1-2b", "name": "LFM2.5-Thinking 1.2B", "level": "model", "family_id": null, "developer": [ "liquid-ai" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "1.2B", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "CPU server", "laptop", "mobile or NPU" ], "footprint_note": "Ollama GGUF build", "runtimes": [ "ollama" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small model listed in the Ollama library for CPU, laptop and phone use.", "verified_on": "2026-10-10", "sources": [ "https://ollama.com/library" ], "unverified": [] }, { "id": "liquid-ai-lfm2-5-vl-3b", "name": "LFM2.5-VL-3B", "level": "model", "family_id": "liquid-ai-lfm2-5", "developer": [ "liquid-ai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": null, "params_total": 3000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 128000, "release_date": "2026-08", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "lfm-1-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "laptop", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/LiquidAI" ], "unverified": [] }, { "id": "luciddreamer", "name": "LucidDreamer", "level": "model", "family_id": null, "developer": [ "seoul-national-university" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-11-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Model scored in the WorldScore paper table.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2504.00983v2", "https://arxiv.org/html/2311.13384", "https://arxiv.org/abs/2311.13384" ], "unverified": [] }, { "id": "luma-ai-genie", "name": "Genie", "level": "model", "family_id": null, "developer": [ "luma-ai" ], "model_class": "3D", "modalities_in": [ "text" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Luma AI 3D service; not researched.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Luma_AI" ], "unverified": [] }, { "id": "luma-ai-photon", "name": "Luma Photon", "level": "model", "family_id": null, "developer": [ "luma-ai" ], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-12", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [ "API", "consumer app" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Luma image generation model with character, style and image references, available in the Luma API and Dream Machine; Photon Flash is its faster, cheaper variant.", "verified_on": "2026-10-11", "sources": [ "https://lumalabs.ai/photon", "https://artificialanalysis.ai/image/leaderboard/text-to-image" ], "unverified": [ "openness" ] }, { "id": "luma-ai-ray-3-2", "name": "Ray 3.2", "level": "model", "family_id": null, "developer": [ "luma-ai" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Luma AI video model in Dream Machine, named in BFL's comparison.", "verified_on": "2026-10-10", "sources": [ "https://datanorth.ai/news/black-forest-labs-releases-flux-3", "https://en.wikipedia.org/wiki/Luma_AI" ], "unverified": [] }, { "id": "luma-ai-uni-1", "name": "Uni-1", "level": "model", "family_id": null, "developer": [ "luma-ai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier Luma image model.", "verified_on": "2026-10-10", "sources": [ "https://arena.ai/leaderboard/text-to-image" ], "unverified": [] }, { "id": "luma-ai-uni-1-1", "name": "Uni-1.1", "level": "model", "family_id": null, "developer": [ "luma-ai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Luma AI proprietary image model (also Uni-1.1-max).", "verified_on": "2026-10-10", "sources": [ "https://arena.ai/leaderboard/text-to-image", "https://en.wikipedia.org/wiki/Luma_AI" ], "unverified": [] }, { "id": "luma-ai-uni-1-1-max", "name": "Uni-1.1-max", "level": "variant", "family_id": "luma-ai-uni-1-1", "developer": [ "luma-ai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Max variant of Luma Uni-1.1.", "verified_on": "2026-10-10", "sources": [ "https://arena.ai/leaderboard/text-to-image" ], "unverified": [] }, { "id": "lumina-dimoo-w-self-grpo", "name": "Lumina-DiMOO w/ Self-GRPO", "level": "model", "family_id": null, "developer": [ "shanghai-ai-lab" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": "Omni diffusion large language model; Self-GRPO is the fourth, self-improving reinforcement learning training stage", "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10-07", "release_date_basis": "inferred from page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Lumina-DiMOO after its Self-GRPO reinforcement learning stage: an 8B omni diffusion language model for multimodal generation and understanding, here tracked for text-to-image generation.", "verified_on": "2026-10-11", "sources": [ "https://arxiv.org/abs/2510.06308" ], "unverified": [] }, { "id": "lunit-insight-cxr", "name": "Lunit INSIGHT CXR", "level": "model", "family_id": null, "developer": [ "lunit" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": "CNN computer-aided detection software as a medical device", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Chest X-ray and TB screening computer-aided detection software from Lunit.", "verified_on": "2026-10-10", "sources": [ "https://www.lunit.io/en/company/news", "https://www.who.int/publications/i/item/9789240022676" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "lunit-insight-mmg", "name": "Lunit INSIGHT MMG", "level": "model", "family_id": null, "developer": [ "lunit" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Mammography cancer detection software from Lunit. Its regulatory status is unverified in the notes.", "verified_on": "2026-10-10", "sources": [ "https://www.lunit.io/en/company/news" ], "unverified": [ "status" ] }, { "id": "lw-detr", "name": "LW-DETR", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": null, "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time DETR used as the base architecture of RF-DETR.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2511.09554v2" ], "unverified": [] }, { "id": "m-a-p-yue-1", "name": "YuE 1", "level": "model", "family_id": null, "developer": [ "m-a-p" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First YuE music model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/m-a-p/YuE2-3B" ], "unverified": [] }, { "id": "m-a-p-yue2-3b", "name": "YuE2-3B", "level": "model", "family_id": null, "developer": [ "m-a-p" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "hybrid", "architecture_detail": "AR-NAR Mixture-of-Transformers plus flow matching plus VAE", "params_total": 3000000000, "params_active": null, "params_checkpoint": null, "params_note": "Named 3B, metadata 4B", "context_length": null, "release_date": "2026-09", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": "Chinese, English tagged; Indonesian not mentioned", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "m-a-p non-commercial open music model with many genre LoRAs on Hugging Face.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/m-a-p/YuE2-3B", "https://huggingface.co/models?pipeline_tag=text-to-audio&sort=trending" ], "unverified": [] }, { "id": "m42-med42-llama3", "name": "Med42 Llama3", "level": "family", "family_id": null, "developer": [ "m42" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-08", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "llama-3-community", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Clinical LLM family from M42 Health (UAE) built on Llama 3.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/m42-health/Llama3-Med42-70B" ], "unverified": [] }, { "id": "m42-med42-llama3-70b", "name": "Llama3-Med42-70B", "level": "model", "family_id": "m42-med42-llama3", "developer": [ "m42" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 70000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-08", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "llama-3-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "70B clinical LLM fine-tuned from Llama3-70B Instruct; MedQA 79.10, MedMCQA 73.20, USMLE 83.80 (developer-reported).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/m42-health/Llama3-Med42-70B" ], "unverified": [] }, { "id": "m42-med42-llama3-8b", "name": "Llama3-Med42-8B", "level": "model", "family_id": "m42-med42-llama3", "developer": [ "m42" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-08", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "llama-3-community", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "8B Med42 clinical LLM; MedQA 62.84 (developer-reported).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/m42-health/Llama3-Med42-70B" ], "unverified": [] }, { "id": "maatouk-llama-3-8b-tele-it", "name": "LLama-3-8B-Tele-it", "level": "model", "family_id": null, "developer": [], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-09-08", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "llama-3-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Telecom domain LLM by Maatouk, Chirino Ampudia, Ying and Tassiulas (Yale affiliation unverified): Llama-3-8B with continued pretraining on telecom data then SFT on Alpaca and Open-instruct.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/AliMaatouk/LLama-3-8B-Tele-it" ], "unverified": [] }, { "id": "mahmood-lab-uni", "name": "UNI", "level": "model", "family_id": null, "developer": [ "mahmood-lab" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": "Self-supervised ViT trained on more than 100M tiles from more than 100k slides; tested on 34 tasks", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-03-19", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download", "research only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Pathology foundation model from the Mahmood Lab at Harvard, with weights for academic research only. Published in Nature Medicine on 19 Mar 2024.", "verified_on": "2026-10-10", "sources": [ "https://www.nature.com/articles/s41591-024-02857-3" ], "unverified": [] }, { "id": "mahmoudashraf-mms-300m-1130-forced-aligner", "name": "mms-300m-1130-forced-aligner", "level": "model", "family_id": "meta-mms", "developer": [ "mahmoudashraf" ], "model_class": "speech recognition", "modalities_in": [ "audio", "text" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "CTC model; MMS-300M checkpoint trained for forced alignment, converted from torchaudio to Hugging Face Transformers", "params_total": 315469471, "params_active": null, "params_checkpoint": null, "params_note": "Hugging Face API Safetensors count", "context_length": null, "release_date": "2024-05-02", "release_date_basis": "registry date", "weights_date": "2024-05-02", "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Card language tags (ISO 639) include Javanese, Sundanese, Malay, Tagalog, Thai and Vietnamese", "indonesian_support": "not stated", "regional_languages": [ "Javanese", "Sundanese" ], "indonesia_hosting": "unknown", "status": null, "summary": "A Hugging Face conversion of the MMS-300M forced-alignment checkpoint, used with the ctc-forced-aligner package to align text to audio with low memory use.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/MahmoudAshraf/mms-300m-1130-forced-aligner", "https://huggingface.co/api/models/MahmoudAshraf/mms-300m-1130-forced-aligner" ], "unverified": [] }, { "id": "mastercard-decision-intelligence-pro", "name": "Decision Intelligence Pro", "level": "model", "family_id": null, "developer": [ "mastercard" ], "model_class": "tabular", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": "Proprietary generative AI; in-house RNN and transformer models referenced; trained on about 125 billion transactions per year", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-02", "release_date_basis": "notes", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Mastercard card fraud model scoring each transaction in about 50 ms; claims average 20% detection uplift, up to 300%.", "verified_on": "2026-10-10", "sources": [ "https://www.cdomagazine.tech/aiml/mastercard-launches-decision-intelligence-pro-new-generative-ai-model-for-fraud-detection" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "matris-10m-oam", "name": "MatRIS-10M-OAM", "level": "model", "family_id": null, "developer": [ "ict-cas" ], "model_class": "materials", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "GNN", "architecture_detail": "Equivariant GNN interatomic potential (MLIP)", "params_total": 10400000, "params_active": null, "params_checkpoint": null, "params_note": "Training data OMat24+MPtrj+sAlex", "context_length": null, "release_date": "2025-10-29", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Machine-learned interatomic potential on Matbench Discovery (rank 5 by CPS, date added 2025-10-29). Developer MatRIS listed as China (unverified).", "verified_on": "2026-10-10", "sources": [ "https://matbench-discovery.materialsproject.org/", "https://github.com/janosh/matbench-discovery/blob/main/models/matris/matris-10m-oam.yml" ], "unverified": [] }, { "id": "meituan-longcat", "name": "LongCat", "level": "family", "family_id": null, "developer": [ "meituan" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "LongCat is the Meituan family of models.", "verified_on": "2026-10-11", "sources": [ "https://arxiv.org/abs/2509.01322", "https://huggingface.co/meituan-longcat/LongCat-2.0" ], "unverified": [] }, { "id": "meituan-longcat-2-0", "name": "LongCat-2.0", "level": "model", "family_id": "meituan-longcat", "developer": [ "meituan" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "LongCat Sparse Attention and N-gram Embedding (135B N-gram embedding parameters); trained on 1M-context data", "params_total": 1600000000000, "params_active": 48000000000, "params_checkpoint": null, "params_note": "About 48B activated per token", "context_length": null, "release_date": "2026-06-29", "release_date_basis": "inferred from page", "weights_date": "2026-07-05", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meituan 1.6 trillion parameter MoE language model with sparse attention for long-horizon coding and agentic tasks.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/meituan-longcat/LongCat-2.0" ], "unverified": [] }, { "id": "meituan-longcat-flash", "name": "LongCat-Flash (LongCat-Flash-Chat)", "level": "model", "family_id": "meituan-longcat", "developer": [ "meituan" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "MoE with zero-computation experts and shortcut-connected MoE (ScMoE)", "params_total": 560000000000, "params_active": 27000000000, "params_checkpoint": null, "params_note": "18.6B to 31.3B active per token, about 27B on average", "context_length": 128000, "release_date": "2025-09-01", "release_date_basis": "inferred from page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download", "consumer app" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meituan 560B non-thinking MoE model with dynamic compute allocation, strong on agentic tasks, with open weights.", "verified_on": "2026-10-11", "sources": [ "https://arxiv.org/abs/2509.01322", "https://huggingface.co/meituan-longcat/LongCat-Flash-Chat" ], "unverified": [] }, { "id": "meituan-longcat-video", "name": "LongCat-Video", "level": "model", "family_id": null, "developer": [ "meituan" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meituan open-weight video model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/models?pipeline_tag=text-to-video&sort=trending" ], "unverified": [] }, { "id": "meituan-mtgr", "name": "MTGR", "level": "model", "family_id": null, "developer": [ "meituan" ], "model_class": "recommendation", "modalities_in": [ "action" ], "modalities_out": [ "action" ], "architecture": "other", "architecture_detail": "Based on HSTU, keeps DLRM cross features, user-level compression", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-05-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "hyperscale cluster" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meituan HSTU-based generative recommender.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2505.18654" ], "unverified": [] }, { "id": "meshy-6-preview", "name": "Meshy 6 Preview", "level": "model", "family_id": null, "developer": [ "meshy" ], "model_class": "3D", "modalities_in": [ "text", "image" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": "Meshy text and image to 3D service with texturing and animation, offered via web, API, MCP and CLI.", "verified_on": "2026-10-10", "sources": [ "https://www.meshy.ai/about", "https://kyodonewsprwire.jp/index.php/release/202511139026" ], "unverified": [] }, { "id": "meta-dinov2", "name": "DINOv2", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": "Self-supervised ViT", "params_total": 304400000, "params_active": null, "params_checkpoint": null, "params_note": "304.4M for dinov2-large; family up to about 1B (unverified)", "context_length": null, "release_date": "2023-07-17", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU", "embedded GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta self-supervised image feature backbone; parent of DINOv3 and backbone of RF-DETR.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/api/models/facebook/dinov2-large", "https://arxiv.org/html/2511.09554v2" ], "unverified": [] }, { "id": "meta-dinov3", "name": "DINOv3", "level": "family", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": "Self-supervised ViT; ViT-7B teacher, distilled ViT-S/S+/B/L/H+ and ConvNeXt T/S/B/L", "params_total": 6716000000, "params_active": null, "params_checkpoint": null, "params_note": "ViT-S 21M to ViT-H+ 840M, ViT-7B 6,716M; ConvNeXt 29M to 198M. Training data LVD-1689M (1.7B images)", "context_length": null, "release_date": "2025-08-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "dinov3-license", "access_routes": [ "open weights download" ], "deployment_tiers": [ "hyperscale cluster", "multi-GPU node", "single datacentre GPU", "CPU server", "embedded GPU" ], "footprint_note": "ViT-7B: GPU cluster or large GPU; distilled S/B and ConvNeXt-T: edge/CPU feasible (judgement)", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta self-supervised vision backbone family for general image features used for classification, detection, segmentation and depth. Task benchmarks not fetched.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/facebook/dinov3-vit7b16-pretrain-sat493m", "https://blockchain.news/ainews/meta-releases-dinov3-advanced-self-supervised-vision-transformer-with-6-7b-parameters-for-superior-image-embeddings" ], "unverified": [] }, { "id": "meta-dinov3-convnext-b", "name": "DINOv3 ConvNeXt-B", "level": "variant", "family_id": "meta-dinov3", "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-08-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "dinov3-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Distilled DINOv3 ConvNeXt-B backbone.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/facebook/dinov3-vit7b16-pretrain-sat493m" ], "unverified": [] }, { "id": "meta-dinov3-convnext-l", "name": "DINOv3 ConvNeXt-L", "level": "variant", "family_id": "meta-dinov3", "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-08-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "dinov3-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Distilled DINOv3 ConvNeXt-L backbone.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/facebook/dinov3-vit7b16-pretrain-sat493m" ], "unverified": [] }, { "id": "meta-dinov3-convnext-s", "name": "DINOv3 ConvNeXt-S", "level": "variant", "family_id": "meta-dinov3", "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-08-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "dinov3-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Distilled DINOv3 ConvNeXt-S backbone.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/facebook/dinov3-vit7b16-pretrain-sat493m" ], "unverified": [] }, { "id": "meta-dinov3-convnext-t", "name": "DINOv3 ConvNeXt-T", "level": "variant", "family_id": "meta-dinov3", "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-08-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "dinov3-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Distilled DINOv3 ConvNeXt-T backbone.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/facebook/dinov3-vit7b16-pretrain-sat493m" ], "unverified": [] }, { "id": "meta-dinov3-vit-7b", "name": "DINOv3 ViT-7B", "level": "variant", "family_id": "meta-dinov3", "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": 6716000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-08-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "dinov3-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ViT-7B teacher of the DINOv3 family.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/facebook/dinov3-vit7b16-pretrain-sat493m" ], "unverified": [] }, { "id": "meta-dinov3-vit-7b-sat493m", "name": "DINOv3 SAT-493M ViT-7B", "level": "variant", "family_id": "meta-dinov3", "developer": [ "meta" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": "Self-supervised ViT on 493M 512x512 satellite images", "params_total": 6700000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "dinov3-license", "access_routes": [ "open weights download" ], "deployment_tiers": [ "hyperscale cluster", "multi-GPU node" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Satellite-imagery branch of DINOv3 (ViT-7B).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/facebook/dinov3-vit7b16-pretrain-sat493m" ], "unverified": [] }, { "id": "meta-dinov3-vit-b", "name": "DINOv3 ViT-B", "level": "variant", "family_id": "meta-dinov3", "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-08-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "dinov3-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Distilled DINOv3 ViT-B backbone.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/facebook/dinov3-vit7b16-pretrain-sat493m" ], "unverified": [] }, { "id": "meta-dinov3-vit-h-plus", "name": "DINOv3 ViT-H+", "level": "variant", "family_id": "meta-dinov3", "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": 840000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-08-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "dinov3-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Distilled DINOv3 ViT-H+ backbone.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/facebook/dinov3-vit7b16-pretrain-sat493m" ], "unverified": [] }, { "id": "meta-dinov3-vit-l", "name": "DINOv3 ViT-L", "level": "variant", "family_id": "meta-dinov3", "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-08-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "dinov3-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Distilled DINOv3 ViT-L backbone.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/facebook/dinov3-vit7b16-pretrain-sat493m" ], "unverified": [] }, { "id": "meta-dinov3-vit-l-sat493m", "name": "DINOv3 SAT-493M ViT-L (distilled)", "level": "variant", "family_id": "meta-dinov3", "developer": [ "meta" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": 300000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "dinov3-license", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Distilled ViT-L satellite branch of DINOv3.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/facebook/dinov3-vit7b16-pretrain-sat493m" ], "unverified": [] }, { "id": "meta-dinov3-vit-s", "name": "DINOv3 ViT-S", "level": "variant", "family_id": "meta-dinov3", "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": 21000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-08-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "dinov3-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Distilled DINOv3 ViT-S backbone.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/facebook/dinov3-vit7b16-pretrain-sat493m" ], "unverified": [] }, { "id": "meta-dinov3-vit-s-plus", "name": "DINOv3 ViT-S+", "level": "variant", "family_id": "meta-dinov3", "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-08-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "dinov3-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Distilled DINOv3 ViT-S+ backbone.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/facebook/dinov3-vit7b16-pretrain-sat493m" ], "unverified": [] }, { "id": "meta-esen-30m-oam", "name": "eSEN-30M-OAM", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "materials", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "GNN", "architecture_detail": "Equivariant GNN interatomic potential (MLIP)", "params_total": 30200000, "params_active": null, "params_checkpoint": null, "params_note": "Training data OMat24+MPtrj+sAlex", "context_length": null, "release_date": "2025-03-17", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Machine-learned interatomic potential on Matbench Discovery (rank 2 by CPS, date added 2025-03-17).", "verified_on": "2026-10-10", "sources": [ "https://matbench-discovery.materialsproject.org/" ], "unverified": [] }, { "id": "meta-esm-2", "name": "ESM-2 / ESMFold", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta protein language model; ancestor of ESM C (unverified).", "verified_on": "2026-10-10", "sources": [ "https://github.com/evolutionaryscale/esm", "https://www.evolutionaryscale.ai/blog/esm3-release" ], "unverified": [ "developer" ] }, { "id": "meta-hstu", "name": "HSTU (Generative Recommenders)", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "recommendation", "modalities_in": [ "action" ], "modalities_out": [ "action" ], "architecture": "dense transformer", "architecture_detail": "Hierarchical Sequential Transduction Unit; recommendation as sequential transduction", "params_total": 1500000000000, "params_active": null, "params_checkpoint": null, "params_note": "Deployed models up to 1.5 trillion parameters", "context_length": null, "release_date": "2024-02-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": null, "access_routes": [ "internal only" ], "deployment_tiers": [ "hyperscale cluster" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta generative recommender; up to 65.8% NDCG over baselines, 5.3x to 15.2x faster than FlashAttention2 transformers at 8,192 length, +12.4% online A/B. Code Apache-2.0.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2402.17152", "https://github.com/facebookresearch/generative-recommenders", "https://github.com/meta-recsys/generative-recommenders" ], "unverified": [] }, { "id": "meta-llama-2", "name": "Llama 2", "level": "family", "family_id": null, "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": "llama-2-community", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "LLM used in OpenVLA.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/indonlp", "https://arxiv.org/abs/2406.09246", "https://huggingface.co/meta-llama/Llama-2-7b" ], "unverified": [] }, { "id": "meta-llama-2-70b", "name": "Llama 2 70B", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Model used as an MLPerf Training v6.0 workload.", "verified_on": "2026-10-10", "sources": [ "https://mlcommons.org/benchmarks/training/" ], "unverified": [] }, { "id": "meta-llama-2-7b", "name": "Llama-2-7B", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 7000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": "llama-2-community", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Base model of Llama Guard.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Yellow-AI-NLP/komodo-7b-base", "https://arxiv.org/abs/2312.06674" ], "unverified": [] }, { "id": "meta-llama-3-1-405b", "name": "Llama 3.1 405B", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Model used as an MLPerf Training v6.0 workload.", "verified_on": "2026-10-10", "sources": [ "https://mlcommons.org/benchmarks/training/" ], "unverified": [] }, { "id": "meta-llama-3-1-405b-instruct", "name": "Llama-3.1-405B-Instruct", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Meta Llama 3.1 405B instruction model, named as the reference model of Llama-3.1-Nemotron-Ultra-253B-v1.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/nvidia/Llama-3_1-Nemotron-Ultra-253B-v1" ], "unverified": [] }, { "id": "meta-llama-3-1-70b", "name": "Llama 3.1 70B", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": 70000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Base of Llama-Sahabat-AI-v2-70B-IT.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/GoToCompany/Llama-Sahabat-AI-v2-70B-IT" ], "unverified": [] }, { "id": "meta-llama-3-1-70b-instruct", "name": "Llama 3.1 70B Instruct", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 70000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": "llama-3-1-community", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/GoToCompany/Llama-Sahabat-AI-v2-70B-IT" ], "unverified": [] }, { "id": "meta-llama-3-1-8b", "name": "Llama 3.1 8B", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Model used as an MLPerf Training v6.0 workload.", "verified_on": "2026-10-10", "sources": [ "https://mlcommons.org/benchmarks/training/", "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B" ], "unverified": [] }, { "id": "meta-llama-3-1-8b-instruct", "name": "Llama-3.1-8B-Instruct", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-07-18", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Most-derived base on HF: 3,257 finetunes, 2,954 adapters, 922 quantizations, 104 merges (2026-10-10).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct" ], "unverified": [] }, { "id": "meta-llama-3-2", "name": "Llama 3.2", "level": "family", "family_id": null, "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-10-24", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": "llama-3-2-community", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta Llama 3.2 family including the 1B and 3B edge models.", "verified_on": "2026-10-10", "sources": [ "https://ai.meta.com/blog/meta-llama-quantized-lightweight-models/" ], "unverified": [ "licence" ] }, { "id": "meta-llama-3-2-1b", "name": "Llama 3.2 1B", "level": "model", "family_id": "meta-llama-3-2", "developer": [ "meta" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 1000000000, "params_active": null, "params_checkpoint": null, "params_note": "1B", "context_length": null, "release_date": "2024-10-24", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": "llama-3-2-community", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Llama 3.2 1B BF16 model, base of Meta's quantized edge versions.", "verified_on": "2026-10-10", "sources": [ "https://ai.meta.com/blog/meta-llama-quantized-lightweight-models/", "https://anthropic.com/research/open-source-circuit-tracing" ], "unverified": [ "licence" ] }, { "id": "meta-llama-3-2-1b-instruct", "name": "Llama-3.2-1B-Instruct", "level": "model", "family_id": "meta-llama-3-2", "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "HF model tree 2026-10-10: 1,812 finetunes, 676 adapters, 426 quantizations, 17 merges", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Llama 3.2 1B instruct model; widely used base on Hugging Face.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct" ], "unverified": [] }, { "id": "meta-llama-3-2-1b-qlora", "name": "Llama 3.2 1B quantized (QLoRA)", "level": "variant", "family_id": "meta-llama-3-2", "developer": [ "meta" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "1B", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": "llama-3-2-community", "access_routes": [], "deployment_tiers": [ "mobile or NPU", "offline capable" ], "footprint_note": "56% smaller on average than BF16; 41% less memory. 4-bit groupwise weights (group 32), 8-bit dynamic activations. 2 to 4x speedup; decode 2.5x, prefill 4.2x on average. Android (OnePlus 12, Samsung S24+, S22) via ExecuTorch", "runtimes": [ "executorch" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta's QLoRA quantized Llama 3.2 1B for Android phones via ExecuTorch.", "verified_on": "2026-10-10", "sources": [ "https://ai.meta.com/blog/meta-llama-quantized-lightweight-models/" ], "unverified": [ "licence" ] }, { "id": "meta-llama-3-2-1b-spinquant", "name": "Llama 3.2 1B quantized (SpinQuant)", "level": "variant", "family_id": "meta-llama-3-2", "developer": [ "meta" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "1B", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": "llama-3-2-community", "access_routes": [], "deployment_tiers": [ "mobile or NPU", "offline capable" ], "footprint_note": "56% smaller on average than BF16; 41% less memory. 4-bit groupwise weights (group 32), 8-bit dynamic activations. 2 to 4x speedup; decode 2.5x, prefill 4.2x on average. Android (OnePlus 12, Samsung S24+, S22) via ExecuTorch", "runtimes": [ "executorch" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta's SpinQuant quantized Llama 3.2 1B for Android phones via ExecuTorch.", "verified_on": "2026-10-10", "sources": [ "https://ai.meta.com/blog/meta-llama-quantized-lightweight-models/" ], "unverified": [ "licence" ] }, { "id": "meta-llama-3-2-3b", "name": "Llama 3.2 3B", "level": "model", "family_id": "meta-llama-3-2", "developer": [ "meta" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "3B", "context_length": null, "release_date": "2024-10-24", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": "llama-3-2-community", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Llama 3.2 3B BF16 model, base of Meta's quantized edge versions.", "verified_on": "2026-10-10", "sources": [ "https://ai.meta.com/blog/meta-llama-quantized-lightweight-models/" ], "unverified": [ "licence" ] }, { "id": "meta-llama-3-2-3b-qlora", "name": "Llama 3.2 3B quantized (QLoRA)", "level": "variant", "family_id": "meta-llama-3-2", "developer": [ "meta" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "3B", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": "llama-3-2-community", "access_routes": [], "deployment_tiers": [ "mobile or NPU", "offline capable" ], "footprint_note": "56% smaller on average than BF16; 41% less memory. 4-bit groupwise weights (group 32), 8-bit dynamic activations. 2 to 4x speedup; decode 2.5x, prefill 4.2x on average. Android (OnePlus 12, Samsung S24+, S22) via ExecuTorch", "runtimes": [ "executorch" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta's QLoRA quantized Llama 3.2 3B for Android phones via ExecuTorch.", "verified_on": "2026-10-10", "sources": [ "https://ai.meta.com/blog/meta-llama-quantized-lightweight-models/" ], "unverified": [ "licence" ] }, { "id": "meta-llama-3-2-3b-spinquant", "name": "Llama 3.2 3B quantized (SpinQuant)", "level": "variant", "family_id": "meta-llama-3-2", "developer": [ "meta" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "3B", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": "llama-3-2-community", "access_routes": [], "deployment_tiers": [ "mobile or NPU", "offline capable" ], "footprint_note": "56% smaller on average than BF16; 41% less memory. 4-bit groupwise weights (group 32), 8-bit dynamic activations. 2 to 4x speedup; decode 2.5x, prefill 4.2x on average. Android (OnePlus 12, Samsung S24+, S22) via ExecuTorch", "runtimes": [ "executorch" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta's SpinQuant quantized Llama 3.2 3B for Android phones via ExecuTorch.", "verified_on": "2026-10-10", "sources": [ "https://ai.meta.com/blog/meta-llama-quantized-lightweight-models/" ], "unverified": [ "licence" ] }, { "id": "meta-llama-3-3-70b-instruct", "name": "Llama-3.3-70B-Instruct", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-11-26", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": "llama-3-3-community", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta model used as the base of DeepSeek-R1-Distill-Llama-70B.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B", "https://huggingface.co/meta-llama/Llama-3.3-70B-Instruct" ], "unverified": [] }, { "id": "meta-llama-3-70b-instruct", "name": "Llama3-70B Instruct", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": 70000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Base model of Med42 Llama3 70B.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/m42-health/Llama3-Med42-70B" ], "unverified": [] }, { "id": "meta-llama-3-8b", "name": "Llama 3 8B", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Base of Sahabat-AI 8B and LLama-3-8B-Tele-it.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/models?search=sahabat", "https://blogs.nvidia.com/?p=75648", "https://huggingface.co/AliMaatouk/LLama-3-8B-Tele-it" ], "unverified": [] }, { "id": "meta-llama-4", "name": "Llama 4", "level": "family", "family_id": null, "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-04", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "llama-4-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta's Llama 4 MoE family; no newer Llama on HF as of 2026-10-10.", "verified_on": "2026-10-10", "sources": [ "https://codersera.com/blog/open-source-llms-landscape-2026/", "https://huggingface.co/meta-llama", "https://arxiv.org/abs/2504.20879" ], "unverified": [] }, { "id": "meta-llama-4-maverick", "name": "Llama 4 Maverick", "level": "model", "family_id": "meta-llama-4", "developer": [ "meta" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "MoE (128 experts)", "params_total": 400000000000, "params_active": 17000000000, "params_checkpoint": null, "params_note": "About 400B total", "context_length": null, "release_date": "2025-04", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "llama-4-community", "access_routes": [ "open weights download" ], "deployment_tiers": [ "multi-GPU node", "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta's Llama 4 MoE model with 128 experts.", "verified_on": "2026-10-10", "sources": [ "https://codersera.com/blog/open-source-llms-landscape-2026/", "https://huggingface.co/meta-llama" ], "unverified": [] }, { "id": "meta-llama-4-scout", "name": "Llama 4 Scout", "level": "model", "family_id": "meta-llama-4", "developer": [ "meta" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "MoE (16 experts)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-04", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "llama-4-community", "access_routes": [ "open weights download" ], "deployment_tiers": [ "multi-GPU node", "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta's Llama 4 MoE model with 16 experts.", "verified_on": "2026-10-10", "sources": [ "https://codersera.com/blog/open-source-llms-landscape-2026/", "https://huggingface.co/meta-llama" ], "unverified": [] }, { "id": "meta-llama-guard", "name": "Llama Guard", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "guard", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-12", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Llama2-7B fine-tuned on a low-volume dataset to classify prompts and responses against a risk taxonomy.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2312.06674" ], "unverified": [] }, { "id": "meta-llama-guard-4-12b", "name": "Llama Guard 4 12B", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "guard", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 12000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-04", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "llama-4-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English + 7 languages (FR, DE, HI, IT, PT, ES, TH); no Indonesian", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta's guard model on MLCommons hazards S1 to S13 plus S14 (code interpreter abuse), for text and multiple images. In-house test set: English F1 61%, multilingual F1 51%.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/meta-llama/Llama-Guard-4-12B" ], "unverified": [] }, { "id": "meta-llama-prompt-guard-2-22m", "name": "Llama Prompt Guard 2 22M", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "guard", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 22000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-04", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "llama-4-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Multilingual (unverified)", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta prompt injection and jailbreak detection model (mDeBERTa base unverified; card not read).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/meta-llama" ], "unverified": [ "licence", "languages_note", "architecture_detail" ] }, { "id": "meta-llama-prompt-guard-2-86m", "name": "Llama Prompt Guard 2 86M", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "guard", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 86000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-04", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "llama-4-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Multilingual (unverified)", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta prompt injection and jailbreak detection model (mDeBERTa base unverified; card not read).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/meta-llama" ], "unverified": [ "licence", "languages_note", "architecture_detail" ] }, { "id": "meta-meta-llama-3-8b-instruct", "name": "Meta-Llama-3-8B-Instruct", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Widely used base on HF: 1,152 finetunes, 1,210 adapters, 278 quantizations, 413 merges (2026-10-10).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct" ], "unverified": [] }, { "id": "meta-mms", "name": "MMS", "level": "family", "family_id": null, "developer": [ "meta" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multilingual speech model lineage; Omnilingual ASR descends from it; fine-tuned for Indonesian in the IDSV study.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2511.09690", "https://arxiv.org/html/2410.08828v2" ], "unverified": [ "developer", "indonesian_measured", "indonesia_hosting" ] }, { "id": "meta-mms-1b-all", "name": "mms-1b-all", "level": "variant", "family_id": "meta-mms", "developer": [ "meta" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-05-27", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MMS model evaluated zero-shot on Indonesian IDSV.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2410.08828v2", "https://huggingface.co/facebook/mms-1b-all" ], "unverified": [ "developer", "indonesia_hosting" ], "indonesian_measured": [ { "benchmark": "IDSV Indonesian speech-variability test set, all groups (mms-1b-all, not fine-tuned)", "score": "29.30", "metric": "WER %, lower is better", "run_by": "paper authors (arXiv 2410.08828)", "url": "https://arxiv.org/html/2410.08828v2" } ] }, { "id": "meta-muse-glimmer", "name": "Muse Glimmer", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Named as an open-weight Meta model by a secondary source (unverified on Meta pages).", "verified_on": "2026-10-10", "sources": [ "https://slash-digital.io/en/insights/ai-models-2026/" ], "unverified": [ "openness" ] }, { "id": "meta-muse-glimmer-30b", "name": "Muse Glimmer 30B", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "Dense, 52 layers, GQA; DFlash speculative-decoding drafter", "params_total": 29600000000, "params_active": null, "params_checkpoint": null, "params_note": "Includes 1.8B perception encoder", "context_length": 131072, "release_date": "2026-08-10", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": "24 GB GPU at about 17 GB K-quant; under 20 GB at 4-bit; 64 GB full precision", "runtimes": [], "languages_note": null, "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta's first Apache-licensed open release, a 30B dense agent model; no Llama 5 found. Adapters 13, finetunes 50, merges 1, quants 177.", "verified_on": "2026-10-10", "sources": [ "https://runtimewire.com/article/meta-open-sources-muse-glimmer-agent-model-under-apache-2-0", "https://huggingface.co/meta-models", "https://huggingface.co/meta-models/Muse-Glimmer-30B", "https://leaderboard.sea-lion.ai/detailed/SEA", "https://leaderboard.sea-lion.ai/detailed/ID" ], "unverified": [ "indonesia_hosting" ], "indonesian_measured": [ { "benchmark": "SEA-HELM Indonesian leaderboard (ID aggregate, first score column of the Indonesian Competency and Task Scores table)", "score": "75.61 (95% CI -1.13 to +1.10)", "metric": "score 0 to 100, average of 2000 bootstraps with 95% CI", "run_by": "AI Singapore (SEA-HELM leaderboard maintainer)", "url": "https://leaderboard.sea-lion.ai/detailed/ID" } ] }, { "id": "meta-muse-image", "name": "Muse Image", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-07", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta Superintelligence Labs image model, Jul 2026.", "verified_on": "2026-10-10", "sources": [ "https://ai.meta.com/blog/", "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text", "https://arena.ai/leaderboard/text-to-image", "https://en.wikipedia.org/wiki/Meta_Superintelligence_Labs" ], "unverified": [] }, { "id": "meta-muse-spark", "name": "Muse Spark", "level": "model", "family_id": "meta-muse-spark-family", "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta model served commercially without weights.", "verified_on": "2026-10-10", "sources": [ "https://ai.meta.com/blog/", "https://arena.ai/leaderboard/chat/vision", "https://codersera.com/blog/open-source-llms-landscape-2026/" ], "unverified": [] }, { "id": "meta-muse-spark-1-1", "name": "Muse Spark 1.1", "level": "model", "family_id": "meta-muse-spark-family", "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-07-09", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Muse Spark 1.1; top entry on Scale SEAL SWE-Bench Pro public leaderboard (developer: Meta).", "verified_on": "2026-10-10", "sources": [ "https://ai.meta.com/blog/", "https://labs.scale.com/leaderboard/swe_bench_pro_public" ], "unverified": [] }, { "id": "meta-muse-spark-1-2", "name": "Muse Spark 1.2", "level": "model", "family_id": "meta-muse-spark-family", "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://ai.meta.com/blog/", "https://arena.ai/leaderboard/chat/vision" ], "unverified": [] }, { "id": "meta-muse-spark-1-3", "name": "Muse Spark 1.3", "level": "model", "family_id": "meta-muse-spark-family", "developer": [ "meta" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": "2026-09-02", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "Weights not public (AA). Meta API.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta's current closed Muse Spark model; text, image and video in, text out.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/models/muse-spark-1-3", "https://artificialanalysis.ai/leaderboards/models", "https://ai.meta.com/blog/", "https://arena.ai/leaderboard/chat/text", "https://arena.ai/leaderboard/chat/vision" ], "unverified": [] }, { "id": "meta-muse-spark-family", "name": "Muse Spark (family)", "level": "family", "family_id": null, "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta's closed Muse Spark model line.", "verified_on": "2026-10-10", "sources": [ "https://ai.meta.com/blog/", "https://artificialanalysis.ai/models/muse-spark-1-3" ], "unverified": [] }, { "id": "meta-muse-video", "name": "Muse Video", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-07-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta Superintelligence Labs video model.", "verified_on": "2026-10-10", "sources": [ "https://ai.meta.com/blog/", "https://arena.ai/leaderboard/text-to-video", "https://en.wikipedia.org/wiki/Meta_Superintelligence_Labs" ], "unverified": [] }, { "id": "meta-musicgen", "name": "MusicGen", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "autoregressive", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "small / medium", "context_length": null, "release_date": "2023", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta open music model, still trending on Hugging Face.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/models?pipeline_tag=text-to-audio&sort=trending", "https://alphasignal.ai/news/artificial-analysis-music-arena-v1-1-crowns-suno-v6-across-1-000-harder-prompts" ], "unverified": [ "architecture" ] }, { "id": "meta-omnilingual-asr", "name": "Omnilingual ASR", "level": "family", "family_id": null, "developer": [ "meta" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "wav2vec 2.0 encoder + CTC head or about 1.2B LLM-style decoder; in-context few-shot for unseen languages; 300M/1B/3B/7B sizes", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "317M to 6.49B encoders", "context_length": null, "release_date": "2025-11-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server", "embedded GPU", "single datacentre GPU" ], "footprint_note": "300M edge/CPU; 7B single GPU", "runtimes": [], "languages_note": "1,600+ languages; average 7.1 CER over 1,570 languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta multilingual ASR family covering 1,600+ languages; code and models released under Apache 2.0, training corpus CC-BY-4.0.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2511.09690", "https://github.com/facebookresearch/omnilingual-asr" ], "unverified": [] }, { "id": "meta-prompt-guard-86m", "name": "Prompt-Guard-86M", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "guard", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 86000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-04", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "llama-4-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Multilingual (unverified)", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta prompt injection and jailbreak detection model (mDeBERTa base unverified; card not read).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/meta-llama" ], "unverified": [ "licence", "languages_note", "architecture_detail" ] }, { "id": "meta-prophet", "name": "Prophet", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta classical forecasting tool, in maintenance mode since v1.4.0 (bug fixes only).", "verified_on": "2026-10-10", "sources": [ "https://www.beri.net/article/timesfm-vs-chronos-vs-moirai-time-series-foundation-model-licence-bigquery-sagemaker-cost-2026" ], "unverified": [] }, { "id": "meta-sam-1", "name": "SAM 1", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": null, "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First Segment Anything model; ancestor of SAM 3.", "verified_on": "2026-10-10", "sources": [ "https://github.com/facebookresearch/sam3" ], "unverified": [] }, { "id": "meta-sam-2", "name": "SAM 2", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image", "video" ], "modalities_out": null, "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier Segment Anything model; its encoder-decoder is inherited by the SAM 3 tracker.", "verified_on": "2026-10-10", "sources": [ "https://github.com/facebookresearch/sam3" ], "unverified": [] }, { "id": "meta-sam-3", "name": "SAM 3", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image", "video", "text" ], "modalities_out": [ "image", "video" ], "architecture": "dense transformer", "architecture_detail": "Shared vision encoder; DETR-based detector conditioned on text, geometry and image exemplars; tracker inherits SAM 2 encoder-decoder; presence token", "params_total": 848000000, "params_active": null, "params_checkpoint": 859900000, "params_note": "848M per repo; HF facebook/sam3 safetensors 859.9M", "context_length": null, "release_date": "2025-11", "release_date_basis": "announcement page", "weights_date": "2025-11-07", "openness": "open weights restricted", "licence": "sam-license", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": "Single GPU (judgement)", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta promptable concept segmentation model for images and video.", "verified_on": "2026-10-10", "sources": [ "https://www.infoq.com/news/2025/11/meta-sam3", "https://github.com/facebookresearch/sam3", "https://huggingface.co/api/models/facebook/sam3" ], "unverified": [] }, { "id": "meta-sam-3-1", "name": "SAM 3.1 (Object Multiplex)", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image", "video", "text" ], "modalities_out": [ "image", "video" ], "architecture": "dense transformer", "architecture_detail": "SAM 3 update with object multiplexing", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-03-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "sam-license", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "SAM 3 update for video multi-object tracking and segmentation.", "verified_on": "2026-10-10", "sources": [ "https://github.com/facebookresearch/sam3", "https://comfyui-wiki.com/en/models/sam/sam3.1" ], "unverified": [ "params_total" ] }, { "id": "meta-sam-3d-objects", "name": "SAM 3D Objects", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "3D", "modalities_in": [ "image" ], "modalities_out": null, "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Derivative of SAM 3.", "verified_on": "2026-10-10", "sources": [ "https://www.infoq.com/news/2025/11/meta-sam3", "https://comfyui-wiki.com/en/models/sam/sam3.1" ], "unverified": [] }, { "id": "meta-v-jepa", "name": "V-JEPA", "level": "family", "family_id": null, "developer": [ "meta" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta joint-embedding predictive video models.", "verified_on": "2026-10-10", "sources": [ "https://ai.meta.com/blog/v-jepa-2-world-model-benchmarks/" ], "unverified": [] }, { "id": "meta-v-jepa-2", "name": "V-JEPA 2", "level": "model", "family_id": "meta-v-jepa", "developer": [ "meta" ], "model_class": "world model", "modalities_in": [ "video", "image" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "Self-supervised joint-embedding predictive video model, pretrained on 1M+ hours of video and images", "params_total": 1200000000, "params_active": null, "params_checkpoint": null, "params_note": "Over 1M hours of video plus 1M images, then 62 hours of robot data.", "context_length": null, "release_date": "2025-06-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta FAIR latent predictive world model; code and checkpoints released for commercial and research use (licence name not stated).", "verified_on": "2026-10-10", "sources": [ "https://ai.meta.com/blog/v-jepa-2-world-model-benchmarks/", "https://arxiv.org/abs/2506.09985" ], "unverified": [] }, { "id": "meta-v-jepa-2-ac", "name": "V-JEPA 2-AC", "level": "model", "family_id": "meta-v-jepa", "developer": [ "meta" ], "model_class": "robot policy", "modalities_in": [ "video", "action" ], "modalities_out": [ "action" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-06-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Action-conditioned V-JEPA 2 post-trained on under 62 h of DROID robot video; zero-shot pick-and-place on Franka arms in 2 labs.", "verified_on": "2026-10-10", "sources": [ "https://ai.meta.com/blog/v-jepa-2-world-model-benchmarks/", "https://arxiv.org/abs/2506.09985" ], "unverified": [] }, { "id": "meta-w2v-bert-2-0", "name": "w2v-bert-2.0", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [], "architecture": "other", "architecture_detail": "Conformer-based W2v-BERT 2.0 speech encoder; bare checkpoint without a task head", "params_total": 600000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-12-19", "release_date_basis": "registry date", "weights_date": "2023-12-19", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Pre-trained on 4.5M hours of unlabelled audio in more than 143 languages; card tags include Indonesian, Javanese, Malay, Tagalog, Thai and Vietnamese", "indonesian_support": "listed by developer", "regional_languages": [ "Javanese" ], "indonesia_hosting": "unknown", "status": null, "summary": "W2v-BERT 2.0 is the self-supervised speech encoder at the core of Meta Seamless models; it needs fine-tuning for ASR or audio classification.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/facebook/w2v-bert-2.0", "https://huggingface.co/api/models/facebook/w2v-bert-2.0" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "meta-wav2vec-2-0", "name": "wav2vec 2.0", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": null, "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Speech encoder lineage from which Omnilingual ASR descends.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2511.09690" ], "unverified": [ "developer" ] }, { "id": "meta-wav2vec2-large-xlsr-53", "name": "wav2vec2-large-xlsr-53", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Multilingual wav2vec 2.0 speech model used as the base of the Indonesian, Javanese and Sundanese ASR model.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/indonesian-nlp/wav2vec2-indonesian-javanese-sundanese" ], "unverified": [] }, { "id": "micronet", "name": "MicroNet", "level": "model", "family_id": null, "developer": [], "model_class": "TinyML", "modalities_in": [ "text" ], "modalities_out": [ "labels and boxes" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "microcontroller", "mobile or NPU" ], "footprint_note": "MicroNet Large INT8 KWS: 658,832 bytes, accuracy 0.965 (definition.yaml)", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Model listed in Arm ML-Zoo, optimised for Arm IP (Cortex-A, Cortex-M, Mali, Ethos-U) in TensorFlow Lite format.", "verified_on": "2026-10-10", "sources": [ "https://github.com/ARM-software/ML-zoo", "https://raw.githubusercontent.com/ARM-software/ML-zoo/HEAD/models/keyword_spotting/micronet_large/tflite_int8/definition.yaml" ], "unverified": [] }, { "id": "microsoft-ai-mai-image", "name": "MAI-Image", "level": "family", "family_id": null, "developer": [ "microsoft" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft AI image model line (MAI-Image-2, 2.5, 2.5-Pro, 2.5-Flash, 2.6, 2.6-Flash).", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text", "https://arena.ai/leaderboard/text-to-image" ], "unverified": [] }, { "id": "microsoft-ai-mai-image-2-5", "name": "MAI-Image-2.5", "level": "model", "family_id": "microsoft-ai-mai-image", "developer": [ "microsoft" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier MAI-Image model.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text" ], "unverified": [] }, { "id": "microsoft-ai-mai-image-2-6", "name": "MAI-Image-2.6", "level": "model", "family_id": "microsoft-ai-mai-image", "developer": [ "microsoft" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-08", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft AI proprietary image model, Aug 2026.", "verified_on": "2026-10-10", "sources": [ "https://microsoft.ai/news/", "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text", "https://arena.ai/leaderboard/text-to-image", "https://en.wikipedia.org/wiki/Microsoft_AI" ], "unverified": [] }, { "id": "microsoft-aurora", "name": "Aurora", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "dense transformer", "architecture_detail": "1.3B atmosphere foundation model; about 11 km", "params_total": 1300000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-05-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "multi-GPU node", "single datacentre GPU" ], "footprint_note": "PyPI microsoft-aurora 2.0.1, 4 Aug 2026; tagged Experimental on Azure", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft Research Earth-system foundation model trained on more than one million hours of geophysical data, with fine-tunes for air quality, waves and cyclones. Code MIT; weight licence not stated on Azure page.", "verified_on": "2026-10-10", "sources": [ "https://labs.ai.azure.com/innovations/aurora/", "https://www.nature.com/articles/s41586-025-09005-y", "https://github.com/microsoft/aurora", "https://pypi.org/pypi/microsoft-aurora/json", "https://huggingface.co/api/models/microsoft/aurora" ], "unverified": [] }, { "id": "microsoft-azure-ai-content-safety", "name": "Azure AI Content Safety", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "guard", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Trained and tested on 8 languages (ZH, EN, FR, DE, ES, IT, JA, PT); Indonesian not listed (\"may work\", test first)", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft's moderation service: 4 harm categories with severity levels, Prompt Shields, groundedness detection (preview), protected material, task adherence and custom categories. F0 and S0 tiers; page updated 1 Oct 2026.", "verified_on": "2026-10-10", "sources": [ "https://learn.microsoft.com/en-us/azure/ai-services/content-safety/overview" ], "unverified": [] }, { "id": "microsoft-e5-mistral-7b", "name": "e5-mistral-7b", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 7000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft E5 embedding model e5-mistral-7b; no 2026 release found.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/intfloat" ], "unverified": [ "licence" ] }, { "id": "microsoft-lightgbm", "name": "LightGBM", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "gradient boosting", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-07-18", "release_date_basis": "registry date", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Latest 4.7.0 on PyPI 18 Jul 2026", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Gradient boosting library; best GBDT on TabArena after tuning and ensembling.", "verified_on": "2026-10-10", "sources": [ "https://github.com/microsoft/LightGBM/releases", "https://pypi.org/pypi/lightgbm/json", "https://github.com/microsoft/LightGBM" ], "unverified": [] }, { "id": "microsoft-mai-code-1-1-flash", "name": "MAI-Code-1.1-Flash", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Lightweight, agentic Microsoft coding model in GitHub Copilot and VS Code; no benchmarks published (page modified 2026-05-31).", "verified_on": "2026-10-10", "sources": [ "https://microsoft.ai/news/", "https://microsoft.ai/models/", "https://docs.github.com/en/copilot/reference/ai-models/supported-models" ], "unverified": [] }, { "id": "microsoft-mai-cyber-1-flash", "name": "MAI-Cyber-1-Flash", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://microsoft.ai/news/" ], "unverified": [] }, { "id": "microsoft-mai-thinking-1", "name": "MAI-Thinking-1", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-05", "release_date_basis": "inferred from page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "Access via Microsoft Foundry and playground.microsoft.ai. Date approximate (May/June 2026).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft AI reasoning model; vendor claims top SWE-Bench Pro results at a mid-weight price (no score shown).", "verified_on": "2026-10-10", "sources": [ "https://microsoft.ai/news/" ], "unverified": [ "release_date" ] }, { "id": "microsoft-mai-voice-2-1", "name": "MAI-Voice-2.1", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "speech synthesis", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://microsoft.ai/news/" ], "unverified": [] }, { "id": "microsoft-maira-2", "name": "MAIRA-2", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "multimodal LLM", "modalities_in": [ "image", "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Frozen image encoder RAD-DINO-MAIRA-2 plus language model Vicuna-7B-v1.5", "params_total": 7000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-06", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "msrla", "access_routes": [ "open weights download", "research only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "7B model that generates grounded chest X-ray findings; Microsoft reported state of the art on MIMIC-CXR findings generation in June 2024. Research use only.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/microsoft/maira-2", "https://www.microsoft.com/en-us/research/publication/maira-2-grounded-radiology-report-generation/" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "microsoft-mdeberta-v3-base", "name": "mDeBERTa-v3-base", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Multilingual encoder pre-trained by Microsoft on CC100 (100 languages).", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/MoritzLaurer/mDeBERTa-v3-base-xnli-multilingual-nli-2mil7" ], "unverified": [] }, { "id": "microsoft-multilingual-e5-large-instruct", "name": "multilingual-e5-large-instruct", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft E5 embedding model multilingual-e5-large-instruct; no 2026 release found.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/intfloat" ], "unverified": [ "licence" ] }, { "id": "microsoft-phi", "name": "Phi", "level": "family", "family_id": null, "developer": [ "microsoft" ], "model_class": "edge LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Microsoft small language model family adapted for industry (Rockwell FT Optix Food & Beverage, Sight Machine Factory Namespace Manager).", "verified_on": "2026-10-10", "sources": [ "https://blogs.microsoft.com/blog/2024/11/13/microsoft-introduces-new-adapted-ai-models-for-industry/" ], "unverified": [ "indonesian_measured" ] }, { "id": "microsoft-phi-3", "name": "Phi-3", "level": "family", "family_id": null, "developer": [ "microsoft" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-04", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Microsoft small model family trained on synthetic plus heavily filtered web data.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2404.14219" ], "unverified": [] }, { "id": "microsoft-phi-3-5-mini", "name": "Phi-3.5-mini", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier Microsoft small model that Phi-4-mini follows.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/microsoft/Phi-4-mini-instruct" ], "unverified": [] }, { "id": "microsoft-phi-3-mini", "name": "Phi-3-mini", "level": "model", "family_id": "microsoft-phi-3", "developer": [ "microsoft" ], "model_class": "edge LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": 3800000000, "params_active": null, "params_checkpoint": null, "params_note": "Trained on 3.3T tokens.", "context_length": null, "release_date": "2024-04", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Deployable on a phone.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "3.8B small language model scoring 69% on MMLU and deployable on a phone.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2404.14219" ], "unverified": [] }, { "id": "microsoft-phi-4", "name": "phi-4", "level": "family", "family_id": null, "developer": [ "microsoft" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft small language model family including Phi-4-mini and Phi-4-mini-flash-reasoning.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/microsoft/Phi-4-reasoning-vision-15B", "https://docs.api.nvidia.com/nim/reference/microsoft-phi-4-mini-flash-reasoning", "https://huggingface.co/microsoft/Phi-4-mini-instruct" ], "unverified": [] }, { "id": "microsoft-phi-4-mini", "name": "Phi-4-mini", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 3800000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-12", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "laptop", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft's 3.8B small model (Dec 2025 refresh).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/microsoft/models?search=phi" ], "unverified": [] }, { "id": "microsoft-phi-4-mini-flash-reasoning", "name": "Phi-4-mini-flash-reasoning", "level": "model", "family_id": "microsoft-phi-4", "developer": [ "microsoft" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "SambaY decoder-hybrid-decoder: state-space layers, Gated Memory Unit and Differential Attention", "params_total": 3852562944, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 64000, "release_date": "2025-06-19", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Up to 10x decoding throughput over Phi-4-mini-reasoning (2K prompt, 32K generation), measured with vLLM on a single A100-80GB (not edge hardware). Quantized footprint and on-device latency unverified", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft 3.85B reasoning model whose card names lightweight deployment on edge or mobile systems as a possible use.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/microsoft/Phi-4-mini-flash-reasoning", "https://huggingface.co/api/models/microsoft/Phi-4-mini-flash-reasoning", "https://docs.api.nvidia.com/nim/reference/microsoft-phi-4-mini-flash-reasoning" ], "unverified": [ "footprint_note" ] }, { "id": "microsoft-phi-4-mini-instruct", "name": "Phi-4-mini-instruct", "level": "model", "family_id": "microsoft-phi-4", "developer": [ "microsoft" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 3836021760, "params_active": null, "params_checkpoint": null, "params_note": "3.8B; safetensors total 3,836,021,760", "context_length": 128000, "release_date": "2025-02-19", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Quantized footprint and on-device latency unverified; LiteRT community conversion exists. Trained on 512 A100-80G. 200,064-token vocabulary", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft 3.8B small model for edge and mobile, MIT licensed, not gated.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/microsoft/Phi-4-mini-instruct", "https://huggingface.co/api/models/microsoft/Phi-4-mini-instruct", "https://docs.api.nvidia.com/nim/reference/microsoft-phi-4-mini-flash-reasoning", "https://www.ertas.ai/blog/phi-4-mini-mobile-ios-android" ], "unverified": [ "footprint_note" ] }, { "id": "microsoft-phi-4-mini-reasoning", "name": "Phi-4-mini-reasoning", "level": "model", "family_id": "microsoft-phi-4", "developer": [ "microsoft" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Phi-4-mini reasoning model; a LiteRT community conversion exists.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/litert-community/Phi-4-mini-reasoning", "https://huggingface.co/microsoft/Phi-4-mini-flash-reasoning" ], "unverified": [] }, { "id": "microsoft-phi-4-multimodal-instruct", "name": "Phi-4-multimodal-instruct", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": 5570000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025", "release_date_basis": "notes", "weights_date": "2025-02-24", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft multimodal LLM listed on the Open ASR Leaderboard.", "verified_on": "2026-10-10", "sources": [ "https://benchmarklist.com/benchmarks/open_asr_leaderboard/", "https://huggingface.co/api/models/microsoft/Phi-4-multimodal-instruct" ], "unverified": [] }, { "id": "microsoft-phi-4-reasoning", "name": "Phi-4-reasoning", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/microsoft/Phi-4-reasoning-vision-15B" ], "unverified": [] }, { "id": "microsoft-phi-4-reasoning-vision-15b", "name": "Phi-4-reasoning-vision-15B", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "Dense, SigLIP-2 vision encoder", "params_total": 15000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 16384, "release_date": "2026-03-04", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft's 15B reasoning model with vision; no Phi-5 found on HF as of 2026-10-10.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/microsoft/Phi-4-reasoning-vision-15B", "https://huggingface.co/microsoft/models?search=phi" ], "unverified": [] }, { "id": "microsoft-prov-gigapath", "name": "Prov-GigaPath", "level": "model", "family_id": null, "developer": [ "microsoft", "providence" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Pathology foundation model attributed to Microsoft and Providence; attribution unverified (page not read).", "verified_on": "2026-10-10", "sources": [ "https://www.nature.com/articles/s41591-024-03141-0", "https://www.nature.com/articles/s41591-024-02857-3" ], "unverified": [ "developer" ] }, { "id": "microsoft-trellis-2-4b", "name": "TRELLIS.2-4B", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "3D", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": "flow matching", "architecture_detail": "Flow-matching transformer plus sparse voxel (O-Voxel) VAE", "params_total": 4000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-12-16", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": "24 GB minimum", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft image-to-mesh model with PBR and GLB export; 1.88M downloads.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/microsoft/TRELLIS.2-4B", "https://huggingface.co/models?pipeline_tag=image-to-3d&sort=trending", "https://en.wikipedia.org/wiki/Microsoft" ], "unverified": [] }, { "id": "microsoft-vibevoice-1-5b", "name": "VibeVoice-1.5B", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open TTS model trending on Hugging Face (2025 to 2026).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/models?pipeline_tag=text-to-speech&sort=trending" ], "unverified": [] }, { "id": "midas", "name": "MiDaS", "level": "model", "family_id": null, "developer": [ "intel", "eth-zurich" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Model listed on PyTorch Hub.", "verified_on": "2026-10-10", "sources": [ "https://pytorch.org/hub/", "https://arxiv.org/html/1907.01341" ], "unverified": [] }, { "id": "midjourney", "name": "Midjourney", "level": "family", "family_id": null, "developer": [ "midjourney" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Midjourney's image generation model versions (V6 to V8.2, Niji).", "verified_on": "2026-10-10", "sources": [ "https://docs.midjourney.com/docs/models", "https://en.wikipedia.org/wiki/Midjourney" ], "unverified": [] }, { "id": "midjourney-niji-7", "name": "Niji 7", "level": "model", "family_id": "midjourney", "developer": [ "midjourney", "spellbrush" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01-09", "release_date_basis": "inferred from page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Midjourney's Niji 7 image model, made with Spellbrush.", "verified_on": "2026-10-10", "sources": [ "https://docs.midjourney.com/docs/models" ], "unverified": [] }, { "id": "midjourney-v6", "name": "Midjourney V6", "level": "model", "family_id": "midjourney", "developer": [ "midjourney" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-12", "release_date_basis": "inferred from page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Midjourney image model version.", "verified_on": "2026-10-10", "sources": [ "https://docs.midjourney.com/docs/models" ], "unverified": [] }, { "id": "midjourney-v6-1", "name": "Midjourney V6.1", "level": "model", "family_id": "midjourney", "developer": [ "midjourney" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Midjourney image model version.", "verified_on": "2026-10-10", "sources": [ "https://docs.midjourney.com/docs/models" ], "unverified": [] }, { "id": "midjourney-v7", "name": "Midjourney V7", "level": "model", "family_id": "midjourney", "developer": [ "midjourney" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-04", "release_date_basis": "inferred from page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Midjourney image model version.", "verified_on": "2026-10-10", "sources": [ "https://docs.midjourney.com/docs/models" ], "unverified": [] }, { "id": "midjourney-v8-0", "name": "Midjourney V8.0 alpha", "level": "model", "family_id": "midjourney", "developer": [ "midjourney" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-03-17", "release_date_basis": "inferred from page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Midjourney image model alpha version.", "verified_on": "2026-10-10", "sources": [ "https://docs.midjourney.com/docs/models" ], "unverified": [] }, { "id": "midjourney-v8-1", "name": "Midjourney V8.1", "level": "model", "family_id": "midjourney", "developer": [ "midjourney" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04-14", "release_date_basis": "inferred from page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Midjourney image model version.", "verified_on": "2026-10-10", "sources": [ "https://docs.midjourney.com/docs/models" ], "unverified": [] }, { "id": "midjourney-v8-2", "name": "Midjourney V8.2", "level": "model", "family_id": "midjourney", "developer": [ "midjourney" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-07-24", "release_date_basis": "inferred from page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app" ], "deployment_tiers": [], "footprint_note": "Cloud only; web app and Discord", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Midjourney's default image model since 24 Jul 2026; adds an Edit Model replacing Omni Reference, Character Reference and Retexture. Midjourney is sued by Disney and Universal (Jun 2025) and Warner Bros. Discovery (Sep 2025).", "verified_on": "2026-10-10", "sources": [ "https://docs.midjourney.com/docs/models", "https://en.wikipedia.org/wiki/Midjourney" ], "unverified": [] }, { "id": "minimax-h3", "name": "MiniMax H3 (Hailuo 3)", "level": "model", "family_id": "minimax-hailuo", "developer": [ "minimax" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video", "audio" ], "architecture": "diffusion", "architecture_detail": "Single-stream DiT; Qwen3-VL-32B text features; 16x/4x VAE; 3 stages (Context-IR, Base 768p, Regenerate-2K)", "params_total": 33000000000, "params_active": null, "params_checkpoint": null, "params_note": "33B Base, dense", "context_length": null, "release_date": "2026-07-31", "release_date_basis": "secondary report", "weights_date": "2026-08-03", "openness": "open weights restricted", "licence": "minimax-h3-community", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [ "multi-GPU node", "consumer GPU" ], "footprint_note": "4 GPUs recommended with SGLang; ComfyUI claims 12 GB with offload", "runtimes": [ "sglang", "comfyui" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MiniMax open-weight video model, top open model on Artificial Analysis, with 32 kHz stereo native audio, 4 to 15 s clips up to 2K. Its licence excludes the US, EU, UK and South Korea.", "verified_on": "2026-10-10", "sources": [ "https://rits.shanghai.nyu.edu/ai/minimax-ships-h3-weights-with-the-us-and-eu-excluded/", "https://artificialanalysis.ai/video/leaderboard/text-to-video", "https://arena.ai/leaderboard/text-to-video", "https://huggingface.co/models?pipeline_tag=text-to-video&sort=trending", "https://en.wikipedia.org/wiki/MiniMax_(company)" ], "unverified": [] }, { "id": "minimax-h3-max", "name": "MiniMax H3 Max", "level": "model", "family_id": "minimax-hailuo", "developer": [ "minimax" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-08", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "API-only Max version of MiniMax H3.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/video/leaderboard/text-to-video", "https://rits.shanghai.nyu.edu/ai/minimax-ships-h3-weights-with-the-us-and-eu-excluded/" ], "unverified": [] }, { "id": "minimax-hailuo", "name": "Hailuo / MiniMax video", "level": "family", "family_id": null, "developer": [ "minimax" ], "model_class": "video generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MiniMax video line: Video-01, Hailuo-02, Hailuo 2.3, H3, H3 Max.", "verified_on": "2026-10-10", "sources": [ "https://amp.kr-asia.com/minimaxs-hailuo-ai-text-to-video-generator-how-does-it-stack-up", "https://www.hedra.com/models/video/minimax/hailuo-23-standard", "https://en.wikipedia.org/wiki/MiniMax_(company)" ], "unverified": [] }, { "id": "minimax-hailuo-02", "name": "Hailuo-02", "level": "model", "family_id": "minimax-hailuo", "developer": [ "minimax" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MiniMax video model including 02 Pro with native 1080p.", "verified_on": "2026-10-10", "sources": [ "https://amp.kr-asia.com/minimaxs-hailuo-ai-text-to-video-generator-how-does-it-stack-up", "https://www.hedra.com/models/video/minimax/hailuo-23-standard" ], "unverified": [] }, { "id": "minimax-hailuo-2-3", "name": "Hailuo 2.3", "level": "model", "family_id": "minimax-hailuo", "developer": [ "minimax" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10-28", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MiniMax video model (Standard, Pro, Fast Pro), 768p/1080p, 6 or 10 s, no native audio per Hedra.", "verified_on": "2026-10-10", "sources": [ "https://www.hedra.com/models/video/minimax/hailuo-23-standard", "https://amp.kr-asia.com/minimaxs-hailuo-ai-text-to-video-generator-how-does-it-stack-up" ], "unverified": [] }, { "id": "minimax-m", "name": "MiniMax-M", "level": "family", "family_id": null, "developer": [ "minimax" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://www.minimax.io/models/text/m3", "https://huggingface.co/MiniMaxAI" ], "unverified": [] }, { "id": "minimax-minimax-m1", "name": "MiniMax-M1", "level": "model", "family_id": "minimax-m", "developer": [ "minimax" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 456000000000, "params_active": 45900000000, "params_checkpoint": null, "params_note": "Final RL stage $534,700 rental on 512 H800 GPUs for about 3 weeks.", "context_length": null, "release_date": "2025-06-16", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MiniMax reasoning model with a published RL-stage cost anchor.", "verified_on": "2026-10-10", "sources": [ "https://www.minimax.io/models/text/m3", "https://huggingface.co/MiniMaxAI", "https://leaderboard.steel.dev/leaderboards/swe-bench-verified.md", "https://arxiv.org/abs/2506.13585" ], "unverified": [] }, { "id": "minimax-minimax-m2", "name": "MiniMax-M2", "level": "model", "family_id": "minimax-m", "developer": [ "minimax" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 229000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://www.minimax.io/models/text/m3", "https://huggingface.co/MiniMaxAI", "https://leaderboard.steel.dev/leaderboards/swe-bench-verified.md" ], "unverified": [] }, { "id": "minimax-minimax-m2-1", "name": "MiniMax-M2.1", "level": "model", "family_id": "minimax-m", "developer": [ "minimax" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://www.minimax.io/models/text/m3", "https://huggingface.co/MiniMaxAI", "https://leaderboard.steel.dev/leaderboards/swe-bench-verified.md" ], "unverified": [] }, { "id": "minimax-minimax-m2-5", "name": "MiniMax-M2.5", "level": "model", "family_id": "minimax-m", "developer": [ "minimax" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-02-12", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://www.minimax.io/models/text/m3", "https://huggingface.co/MiniMaxAI", "https://leaderboard.steel.dev/leaderboards/swe-bench-verified.md", "https://huggingface.co/MiniMaxAI/MiniMax-M2.5" ], "unverified": [] }, { "id": "minimax-minimax-m2-7", "name": "MiniMax-M2.7", "level": "model", "family_id": "minimax-m", "developer": [ "minimax" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://www.minimax.io/models/text/m3", "https://huggingface.co/MiniMaxAI", "https://leaderboard.steel.dev/leaderboards/swe-bench-verified.md" ], "unverified": [] }, { "id": "minimax-minimax-m3", "name": "MiniMax-M3", "level": "model", "family_id": "minimax-m", "developer": [ "minimax" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "MoE with MiniMax Sparse Attention", "params_total": 428000000000, "params_active": 23000000000, "params_checkpoint": null, "params_note": "About 428B / 23B", "context_length": 1000000, "release_date": "2026-06-01", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights restricted", "licence": "minimax-m3-license", "access_routes": [ "open weights download" ], "deployment_tiers": [ "multi-GPU node" ], "footprint_note": "1M context (512K guaranteed). Estimate, unverified. Announced as open-weight, weights not yet public; the coding notes record minimax-community licence on HF.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MiniMax's natively multimodal MoE model, successor to M2.x.", "verified_on": "2026-10-10", "sources": [ "https://www.minimax.io/models/text/m3", "https://huggingface.co/MiniMaxAI/MiniMax-M3", "https://artificialanalysis.ai/leaderboards/models", "https://arena.ai/leaderboard/chat/text", "https://huggingface.co/MiniMaxAI/MiniMax-M3/blob/main/LICENSE", "https://news.aibase.com/tw/news/28491", "https://huggingface.co/api/models/MiniMaxAI/MiniMax-M3" ], "unverified": [ "deployment_tiers" ] }, { "id": "minimax-music-3", "name": "MiniMax Music 3", "level": "model", "family_id": null, "developer": [ "minimax" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "hybrid", "architecture_detail": "Hierarchical AR Hybrid-LM (Global 8B from Qwen3-8B plus Local 0.6B) plus flow matching plus Flow-VAE", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "About 11B across components", "context_length": null, "release_date": "2026-08-07", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": "CUDA required", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MiniMax open-weight music model, up to 5 min at 32 kHz stereo; top open-weights model and 1000 anchor on AA Music Arena v1.1. Licence file not read.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/MiniMaxAI/MiniMax-Music3", "https://huggingface.co/models?pipeline_tag=text-to-audio&sort=trending", "https://alphasignal.ai/news/artificial-analysis-music-arena-v1-1-crowns-suno-v6-across-1-000-harder-prompts" ], "unverified": [ "licence" ] }, { "id": "minimax-speech-2-8-hd", "name": "Speech 2.8 HD", "level": "model", "family_id": null, "developer": [ "minimax" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-02", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "40-language list includes Indonesian", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MiniMax TTS model (speech-2.8-hd), 19th on the Artificial Analysis TTS arena.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/text-to-speech/leaderboard", "https://www.minimax.io/about", "https://platform.minimax.io/docs/guides/speech-t2a-websocket" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "minimax-speech-2-8-turbo", "name": "Speech 2.8 Turbo", "level": "model", "family_id": null, "developer": [ "minimax" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MiniMax TTS model (speech-2.8-turbo).", "verified_on": "2026-10-10", "sources": [ "https://platform.minimax.io/docs/guides/speech-t2a-websocket" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "minimax-video-01", "name": "Video-01", "level": "model", "family_id": "minimax-hailuo", "developer": [ "minimax" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-09", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MiniMax's first video model, sold as Hailuo AI, Sep 2024.", "verified_on": "2026-10-10", "sources": [ "https://amp.kr-asia.com/minimaxs-hailuo-ai-text-to-video-generator-how-does-it-stack-up" ], "unverified": [] }, { "id": "mistral-ai-codestral", "name": "Codestral", "level": "family", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "code", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://docs.mistral.ai/getting-started/models/models_overview/", "https://docs.mistral.ai/getting-started/models/models_overview/" ], "unverified": [] }, { "id": "mistral-ai-codestral-25-08", "name": "Codestral 25.08", "level": "model", "family_id": "mistral-ai-codestral", "developer": [ "mistral-ai" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-07-30", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "Premier (commercial); cloud, VPC, on-prem (T1/T3).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Mistral's fill-in-the-middle completion model; still Mistral's only current Code model.", "verified_on": "2026-10-10", "sources": [ "https://mistral.ai/news/codestral-25-08", "https://docs.mistral.ai/getting-started/models/models_overview/" ], "unverified": [] }, { "id": "mistral-ai-devstral", "name": "Devstral", "level": "family", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "code", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/mistralai/Devstral-2-123B-Instruct-2512", "https://huggingface.co/mistralai/Devstral-Small-2-24B-Instruct-2512", "https://mistral.ai/news/codestral-25-08" ], "unverified": [] }, { "id": "mistral-ai-devstral-2-123b-instruct-2512", "name": "Devstral 2 (123B, 2512)", "level": "model", "family_id": "mistral-ai-devstral", "developer": [ "mistral-ai" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 123000000000, "params_active": null, "params_checkpoint": null, "params_note": "125B listed", "context_length": 256000, "release_date": "2025-12", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "mistral-modified-mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "multi-GPU node" ], "footprint_note": "T3 estimate (unverified). Deprecated on Mistral API as of Oct 2026.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "Mistral open coding model; API version deprecated.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/mistralai/Devstral-2-123B-Instruct-2512", "https://docs.mistral.ai/getting-started/models/models_overview/" ], "unverified": [] }, { "id": "mistral-ai-devstral-medium-1-0", "name": "Devstral Medium 1.0", "level": "model", "family_id": "mistral-ai-devstral", "developer": [ "mistral-ai" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-07", "release_date_basis": "notes", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://mistral.ai/news/codestral-25-08" ], "unverified": [] }, { "id": "mistral-ai-devstral-small-1-1", "name": "Devstral Small 1.1", "level": "model", "family_id": "mistral-ai-devstral", "developer": [ "mistral-ai" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 24000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-07", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://mistral.ai/news/codestral-25-08" ], "unverified": [] }, { "id": "mistral-ai-devstral-small-2-24b-instruct-2512", "name": "Devstral Small 2 (24B, 2512)", "level": "model", "family_id": "mistral-ai-devstral", "developer": [ "mistral-ai" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 24000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 256000, "release_date": "2025-12", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU", "laptop" ], "footprint_note": "Vendor states single RTX 4090 or 32 GB Mac (T4). API deprecated.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "Mistral 24B open coding model fine-tuned from Mistral-Small-3.1-24B-Base-2503.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/mistralai/Devstral-Small-2-24B-Instruct-2512", "https://huggingface.co/mistralai/Devstral-2-123B-Instruct-2512", "https://docs.mistral.ai/getting-started/models/models_overview/" ], "unverified": [] }, { "id": "mistral-ai-leanstral-1-5-119b-a6b", "name": "Leanstral-1.5-119B-A6B", "level": "model", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 119000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/mistralai/Mistral-Small-4-119B-2603" ], "unverified": [] }, { "id": "mistral-ai-magistral", "name": "Magistral", "level": "family", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "reasoning", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-06-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Mistral AI reasoning model family released June 2025.", "verified_on": "2026-10-10", "sources": [ "https://mistral.ai/news/magistral" ], "unverified": [] }, { "id": "mistral-ai-magistral-medium-1-2", "name": "Magistral Medium 1.2", "level": "model", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://docs.mistral.ai/getting-started/models/models_overview/" ], "unverified": [] }, { "id": "mistral-ai-magistral-small", "name": "Magistral Small", "level": "model", "family_id": "mistral-ai-magistral", "developer": [ "mistral-ai" ], "model_class": "reasoning", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": 24000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-06-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "24B open-weight Magistral reasoning model under Apache 2.0.", "verified_on": "2026-10-10", "sources": [ "https://mistral.ai/news/magistral" ], "unverified": [] }, { "id": "mistral-ai-magistral-small-1-2", "name": "Magistral Small 1.2", "level": "model", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://docs.mistral.ai/getting-started/models/models_overview/" ], "unverified": [] }, { "id": "mistral-ai-mamba-codestral-7b-v0-1", "name": "Mamba-Codestral-7B-v0.1", "level": "model", "family_id": "mistral-ai-codestral", "developer": [ "mistral-ai" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "SSM", "architecture_detail": null, "params_total": 7000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "gated preview" ], "deployment_tiers": [ "laptop" ], "footprint_note": "Gated repo. T5 estimate.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/mistralai" ], "unverified": [ "release_date", "deployment_tiers" ] }, { "id": "mistral-ai-ministral-3-3b-base-2512", "name": "Ministral-3-3B-Base-2512", "level": "model", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 3000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/mistralai/Shieldstral-1.0-3B" ], "unverified": [] }, { "id": "mistral-ai-ministral-3-8b-instruct-2512", "name": "Ministral-3-8B-Instruct-2512", "level": "model", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/nvidia/Nemotron-3-Embed-8B-BF16" ], "unverified": [] }, { "id": "mistral-ai-mistral-7b", "name": "Mistral 7B", "level": "model", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 7000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/intfloat" ], "unverified": [] }, { "id": "mistral-ai-mistral-7b-v0-1", "name": "Mistral-7B-v0.1", "level": "model", "family_id": null, "developer": [], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Widely used base on HF: 949 finetunes, 2,465 adapters, 195 quantizations, 390 merges (2026-10-10).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct" ], "unverified": [] }, { "id": "mistral-ai-mistral-7b-v0-3", "name": "Mistral-7B-v0.3", "level": "model", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 7000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/allenai/wildguard" ], "unverified": [] }, { "id": "mistral-ai-mistral-embed", "name": "Mistral Embed", "level": "model", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "embedding", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Mistral premier (API-only) model.", "verified_on": "2026-10-10", "sources": [ "https://docs.mistral.ai/getting-started/models/models_overview/" ], "unverified": [] }, { "id": "mistral-ai-mistral-large-3", "name": "Mistral Large 3 (2512)", "level": "model", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "MoE with 2.5B vision encoder", "params_total": 675000000000, "params_active": 41000000000, "params_checkpoint": null, "params_note": "Card also says 39B active", "context_length": 256000, "release_date": "2025-12", "release_date_basis": "inferred from page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "multi-GPU node" ], "footprint_note": "NVFP4 and Eagle variants", "runtimes": [], "languages_note": "11 languages listed; no Indonesian", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Mistral AI's 675B MoE text and image model under Apache 2.0. Finetunes 13, quants 7.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/mistralai", "https://huggingface.co/mistralai/Mistral-Large-3-675B-Instruct-2512" ], "unverified": [] }, { "id": "mistral-ai-mistral-large-4", "name": "Mistral Large 4 (v26.10)", "level": "model", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 1050000000000, "params_active": 52000000000, "params_checkpoint": null, "params_note": "News post rounds to 1 trillion. Context 1M as Mistral's own model page states (re-read 15 Oct 2026); Artificial Analysis listed 524k for the Preview, the developer's figure governs.", "context_length": 1000000, "release_date": "2026-10-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API", "gated preview" ], "deployment_tiers": [], "footprint_note": "Weights announced for end of October 2026, not released as of 10 Oct 2026; AA labels it proprietary.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": "Mistral AI's natively multimodal flagship in public preview; announced as open-weight with weights still to come.", "verified_on": "2026-10-10", "sources": [ "https://mistral.ai/news/mistral-large-4/", "https://docs.mistral.ai/models/mistral-large-4-0", "https://docs.mistral.ai/getting-started/models/models_overview/", "https://artificialanalysis.ai/models/mistral-large-4", "https://arena.ai/leaderboard/chat/text", "https://llm-stats.com/ai-news" ], "unverified": [] }, { "id": "mistral-ai-mistral-large-feb-24", "name": "Mistral Large (Feb '24)", "level": "model", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 32000, "release_date": "2024-02-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "consumer app", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": "La Plateforme (hosted in Europe), Azure AI Studio and Azure Machine Learning; self-deployment with weights offered for sensitive use cases", "runtimes": [], "languages_note": "Natively fluent in English, French, Spanish, German and Italian", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Mistral Large is the February 2024 Mistral flagship with native function calling, which Mistral described as the second-ranked model generally available through an API.", "verified_on": "2026-10-11", "sources": [ "https://mistral.ai/news/mistral-large/" ], "unverified": [] }, { "id": "mistral-ai-mistral-large-instruct-2407", "name": "Mistral-Large-Instruct-2407 (Mistral Large 2)", "level": "model", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Mistral Large 2 text model on which Pixtral Large is built.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/mistralai/Pixtral-Large-Instruct-2411" ], "unverified": [] }, { "id": "mistral-ai-mistral-medium", "name": "Mistral Medium", "level": "family", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "multimodal LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Mistral Medium is the Mistral AI line of mid-size enterprise models served through its API.", "verified_on": "2026-10-11", "sources": [ "https://mistral.ai/news/mistral-medium-3", "https://docs.mistral.ai/getting-started/models/models_overview" ], "unverified": [] }, { "id": "mistral-ai-mistral-medium-3", "name": "Mistral Medium 3 (mistral-medium-2505)", "level": "model", "family_id": "mistral-ai-mistral-medium", "developer": [ "mistral-ai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 131072, "release_date": "2025-05-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [ "API", "cloud marketplace" ], "deployment_tiers": [ "multi-GPU node" ], "footprint_note": "Announcement says it can be self-hosted on environments of four GPUs and above", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "Mistral AI enterprise model for coding and multimodal understanding, offered through La Plateforme and cloud marketplaces; Mistral docs list it as deprecated on 22 May 2026 with retirement on 31 August 2026.", "verified_on": "2026-10-11", "sources": [ "https://mistral.ai/news/mistral-medium-3", "https://docs.mistral.ai/getting-started/models/models_overview", "https://openrouter.ai/api/v1/models" ], "unverified": [ "context_length" ] }, { "id": "mistral-ai-mistral-medium-3-1", "name": "Mistral Medium 3.1 (mistral-medium-2508)", "level": "model", "family_id": "mistral-ai-mistral-medium", "developer": [ "mistral-ai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 131072, "release_date": "2025-08-13", "release_date_basis": "registry date", "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "Updated version of Mistral Medium 3 (version 25.08); Mistral docs list it as deprecated on 22 May 2026 with retirement on 31 August 2026.", "verified_on": "2026-10-11", "sources": [ "https://openrouter.ai/api/v1/models", "https://openrouter.ai/mistralai/mistral-medium-3.1", "https://docs.mistral.ai/getting-started/models/models_overview" ], "unverified": [ "context_length" ] }, { "id": "mistral-ai-mistral-medium-3-5", "name": "Mistral Medium 3.5 (v26.04)", "level": "model", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "Dense, merged instruct + reasoning + code, own vision encoder", "params_total": 128000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 256000, "release_date": "2026-04-28", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "mistral-modified-mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "multi-GPU node", "single datacentre GPU" ], "footprint_note": "T3 estimate (unverified).", "runtimes": [], "languages_note": null, "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Mistral model optimised for agentic and coding use; absorbs the Devstral role.", "verified_on": "2026-10-10", "sources": [ "https://docs.mistral.ai/getting-started/models/models_overview/", "https://artificialanalysis.ai/leaderboards/models", "https://arena.ai/leaderboard/chat/text", "https://huggingface.co/mistralai/Mistral-Medium-3.5-128B", "https://codersera.com/blog/open-source-llms-landscape-2026/", "https://leaderboard.sea-lion.ai/detailed/SEA", "https://leaderboard.sea-lion.ai/detailed/ID", "https://docs.mistral.ai/getting-started/changelog/", "https://huggingface.co/api/models/mistralai/Mistral-Medium-3.5-128B", "https://mistral.ai/news", "https://huggingface.co/mistralai/Mistral-Medium-3.5-128B/raw/main/LICENSE", "https://huggingface.co/mistralai/Devstral-2-123B-Instruct-2512/raw/main/LICENSE" ], "unverified": [ "deployment_tiers", "indonesia_hosting" ], "indonesian_measured": [ { "benchmark": "SEA-HELM Indonesian leaderboard (ID aggregate, first score column of the Indonesian Competency and Task Scores table)", "score": "70.09 (95% CI -1.18 to +1.13)", "metric": "score 0 to 100, average of 2000 bootstraps with 95% CI", "run_by": "AI Singapore (SEA-HELM leaderboard maintainer)", "url": "https://leaderboard.sea-lion.ai/detailed/ID" } ] }, { "id": "mistral-ai-mistral-medium-dec-23", "name": "Mistral Medium (Dec '23)", "level": "model", "family_id": "mistral-ai-mistral-medium", "developer": [ "mistral-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-12-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English, French, Italian, German, Spanish and code", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Mistral Medium (mistral-medium) was the highest-quality endpoint of the La Plateforme API beta, serving a prototype model.", "verified_on": "2026-10-11", "sources": [ "https://mistral.ai/news/la-plateforme/" ], "unverified": [] }, { "id": "mistral-ai-mistral-moderation-2", "name": "Mistral Moderation 2", "level": "model", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "guard", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Mistral premier (API-only) model.", "verified_on": "2026-10-10", "sources": [ "https://docs.mistral.ai/getting-started/models/models_overview/" ], "unverified": [] }, { "id": "mistral-ai-mistral-ocr-4-1", "name": "Mistral OCR 4.1", "level": "model", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "OCR and document", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Mistral premier (API-only) model.", "verified_on": "2026-10-10", "sources": [ "https://docs.mistral.ai/getting-started/models/models_overview/" ], "unverified": [] }, { "id": "mistral-ai-mistral-small", "name": "Mistral Small", "level": "family", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Mistral Small is the Mistral AI line of small, low-latency models.", "verified_on": "2026-10-11", "sources": [ "https://mistral.ai/news/mistral-small-3", "https://docs.mistral.ai/getting-started/models/models_overview" ], "unverified": [] }, { "id": "mistral-ai-mistral-small-24b-base-2501", "name": "Mistral-Small-24B-Base-2501", "level": "model", "family_id": "mistral-ai-mistral-small", "developer": [ "mistral-ai" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Pretrained base checkpoint released with Mistral Small 3.", "verified_on": "2026-10-11", "sources": [ "https://mistral.ai/news/mistral-small-3", "https://huggingface.co/api/models/mistralai/Mistral-Small-24B-Instruct-2501" ], "unverified": [] }, { "id": "mistral-ai-mistral-small-3", "name": "Mistral Small 3 (Mistral-Small-24B-Instruct-2501)", "level": "model", "family_id": "mistral-ai-mistral-small", "developer": [ "mistral-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 24000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 32768, "release_date": "2025-01-30", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "API", "open weights download", "cloud marketplace" ], "deployment_tiers": [ "consumer GPU", "laptop" ], "footprint_note": "Announcement: when quantized, runs on a single RTX 4090 or a MacBook with 32GB RAM", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "24B low-latency Mistral AI model released under Apache 2.0; the Mistral API version (mistral-small-2501) was deprecated on 6 November 2025 and retired on 30 November 2025.", "verified_on": "2026-10-11", "sources": [ "https://mistral.ai/news/mistral-small-3", "https://docs.mistral.ai/getting-started/models/models_overview", "https://openrouter.ai/api/v1/models" ], "unverified": [ "context_length" ] }, { "id": "mistral-ai-mistral-small-3-1-24b-base-2503", "name": "Mistral-Small-3.1-24B-Base-2503", "level": "model", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 24000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/mistralai/Devstral-Small-2-24B-Instruct-2512" ], "unverified": [] }, { "id": "mistral-ai-mistral-small-4", "name": "Mistral Small 4", "level": "model", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "MoE (128 experts, 4 active)", "params_total": 119000000000, "params_active": 6500000000, "params_checkpoint": null, "params_note": null, "context_length": 256000, "release_date": "2026-03", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "multi-GPU node", "single datacentre GPU" ], "footprint_note": "T3 estimate (unverified).", "runtimes": [], "languages_note": "Indonesian not named on card", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Mistral model that unifies instruct, reasoning and coding.", "verified_on": "2026-10-10", "sources": [ "https://docs.mistral.ai/getting-started/models/models_overview/", "https://huggingface.co/mistralai", "https://huggingface.co/mistralai/Mistral-Small-4-119B-2603" ], "unverified": [ "deployment_tiers" ] }, { "id": "mistral-ai-mistral-small-sep-24", "name": "Mistral Small (Sep '24)", "level": "model", "family_id": "mistral-ai-mistral-small", "developer": [ "mistral-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 22000000000, "params_active": null, "params_checkpoint": null, "params_note": "Card states 22B; Hugging Face API counts 22,247,282,688", "context_length": 32000, "release_date": "2024-09-17", "release_date_basis": "announcement page", "weights_date": "2024-09-17", "openness": "open weights restricted", "licence": "mistral-research-license-0-1", "access_routes": [ "API", "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": "Card: a single GPU needs at least 44 GB of GPU RAM", "runtimes": [ "vllm" ], "languages_note": "Card language tags: en, fr, de, es, it, pt, zh, ja, ru, ko", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Mistral Small v24.09 (Mistral-Small-Instruct-2409) is a 22B instruct model with function calling, under the Mistral Research License for self-deployment and on the Mistral API for commercial use.", "verified_on": "2026-10-11", "sources": [ "https://mistral.ai/news/september-24-release/", "https://huggingface.co/mistralai/Mistral-Small-Instruct-2409", "https://huggingface.co/api/models/mistralai/Mistral-Small-Instruct-2409" ], "unverified": [] }, { "id": "mistral-ai-mixtral-8x22b-instruct", "name": "Mixtral 8x22B Instruct", "level": "model", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "Sparse mixture of experts", "params_total": 141000000000, "params_active": 39000000000, "params_checkpoint": null, "params_note": null, "context_length": 64000, "release_date": "2024-04-17", "release_date_basis": "announcement page", "weights_date": "2024-04-16", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "API", "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [ "vllm" ], "languages_note": "English, French, Italian, German and Spanish", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Mixtral 8x22B Instruct is the instruction-tuned Mistral sparse mixture-of-experts model with 39B active of 141B parameters, native function calling and a 64K context.", "verified_on": "2026-10-11", "sources": [ "https://mistral.ai/news/mixtral-8x22b/", "https://huggingface.co/mistralai/Mixtral-8x22B-Instruct-v0.1", "https://huggingface.co/api/models/mistralai/Mixtral-8x22B-Instruct-v0.1" ], "unverified": [] }, { "id": "mistral-ai-mixtral-8x22b-v0-1", "name": "Mixtral-8x22B-v0.1", "level": "model", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": "MoE", "architecture_detail": null, "params_total": 141000000000, "params_active": 39000000000, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Base sparse mixture-of-experts model of Mixtral 8x22B Instruct.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/mistralai/Mixtral-8x22B-Instruct-v0.1", "https://mistral.ai/news/mixtral-8x22b/" ], "unverified": [] }, { "id": "mistral-ai-pixtral", "name": "Pixtral", "level": "family", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "multimodal LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Mistral AI multimodal model line that includes Pixtral 12B and Pixtral Large.", "verified_on": "2026-10-11", "sources": [ "https://mistral.ai/news/pixtral-large/", "https://huggingface.co/mistralai/Pixtral-Large-Instruct-2411" ], "unverified": [] }, { "id": "mistral-ai-pixtral-large", "name": "Pixtral Large", "level": "model", "family_id": "mistral-ai-pixtral", "developer": [ "mistral-ai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "123B multimodal decoder plus 1B vision encoder, built on Mistral Large 2", "params_total": 124000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 128000, "release_date": "2024-11-18", "release_date_basis": "announcement page", "weights_date": "2024-11-18", "openness": "open weights restricted", "licence": "mistral-research-license-0-1", "access_routes": [ "API", "open weights download" ], "deployment_tiers": [ "multi-GPU node" ], "footprint_note": "vLLM serving example uses tensor-parallel-size 8; commercial use needs a separate Mistral commercial licence", "runtimes": [ "vllm" ], "languages_note": "Hugging Face language tags: en, fr, de, es, it, pt, zh, ja, ru, ko", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Mistral AI 124B multimodal model that understands documents, charts and natural images while keeping Mistral Large 2 text performance.", "verified_on": "2026-10-11", "sources": [ "https://mistral.ai/news/pixtral-large/", "https://huggingface.co/mistralai/Pixtral-Large-Instruct-2411", "https://mistral.ai/licenses/MRL-0.1.md" ], "unverified": [] }, { "id": "mistral-ai-shieldstral-1-0-3b", "name": "Shieldstral 1.0 3B", "level": "model", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "guard", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 3000000000, "params_active": null, "params_checkpoint": null, "params_note": "4B listed", "context_length": null, "release_date": "2026-07-28", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "12 languages; no Indonesian", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Mistral AI's guard model that takes a policy at inference as a natural-language yes/no query (no fixed taxonomy), for text and image. HF 5 Aug 2026.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/mistralai/Shieldstral-1.0-3B" ], "unverified": [] }, { "id": "mistral-ai-voxtral-mini-transcribe-2", "name": "Voxtral Mini Transcribe 2", "level": "model", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "speech recognition", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Mistral premier (API-only) model.", "verified_on": "2026-10-10", "sources": [ "https://docs.mistral.ai/getting-started/models/models_overview/" ], "unverified": [] }, { "id": "mistral-ai-voxtral-small", "name": "Voxtral Small", "level": "model", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": 24000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Mistral speech model (24B); Open ASR average WER 6.62.", "verified_on": "2026-10-10", "sources": [ "https://select.keenable.ai/r/speech-recognition-accuracy-leaders-2022-2026/JbURQjlVooLQ6J10NkbRVw" ], "unverified": [] }, { "id": "mit-han-lab-mcunet", "name": "MCUNet", "level": "family", "family_id": null, "developer": [ "mit-han-lab" ], "model_class": "TinyML", "modalities_in": [ "text" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": "TinyNAS neural architecture search co-designed with TinyEngine inference engine", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [ "microcontroller" ], "footprint_note": "ImageNet models: SRAM 242 to 456 kB, Flash 878 to 1876 kB; profiled on STM32F746", "runtimes": [ "tinyengine" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MIT Han Lab system for running neural networks on microcontrollers; last dated news Mar 2024 (low activity).", "verified_on": "2026-10-10", "sources": [ "https://github.com/mit-han-lab/mcunet" ], "unverified": [] }, { "id": "mit-han-lab-mcunet-in4", "name": "mcunet-in4", "level": "model", "family_id": "mit-han-lab-mcunet", "developer": [ "mit-han-lab" ], "model_class": "TinyML", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [ "microcontroller" ], "footprint_note": "int8", "runtimes": [ "tinyengine" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MCUNet ImageNet microcontroller model.", "verified_on": "2026-10-10", "sources": [ "https://github.com/mit-han-lab/mcunet" ], "unverified": [] }, { "id": "mit-han-lab-mcunet-vww2", "name": "mcunet-vww2", "level": "model", "family_id": "mit-han-lab-mcunet", "developer": [ "mit-han-lab" ], "model_class": "TinyML", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [ "microcontroller" ], "footprint_note": "int8", "runtimes": [ "tinyengine" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MCUNet visual wake words microcontroller model.", "verified_on": "2026-10-10", "sources": [ "https://github.com/mit-han-lab/mcunet" ], "unverified": [] }, { "id": "mlcommons-mlperf-tiny-1d-ds-cnn-wakeword", "name": "MLPerf Tiny 1D DS-CNN (streaming wakeword)", "level": "model", "family_id": null, "developer": [ "mlcommons" ], "model_class": "TinyML", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [ "microcontroller" ], "footprint_note": "MCU-class", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MLPerf Tiny reference model for streaming wakeword on microcontrollers and NPUs.", "verified_on": "2026-10-10", "sources": [ "https://mlcommons.org/benchmarks/inference-tiny/", "https://www.globenewswire.com/news-release/2025/09/17/3151817/0/en/MLCommons-New-MLPerf-Tiny-v1-3-Benchmark-Results-Released.html" ], "unverified": [ "licence" ] }, { "id": "mlcommons-mlperf-tiny-dense-ae-toycar", "name": "MLPerf Tiny Dense AE (ToyCar)", "level": "model", "family_id": null, "developer": [ "mlcommons" ], "model_class": "TinyML", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [ "microcontroller" ], "footprint_note": "MCU-class", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MLPerf Tiny reference model for anomaly detection on microcontrollers and NPUs.", "verified_on": "2026-10-10", "sources": [ "https://mlcommons.org/benchmarks/inference-tiny/" ], "unverified": [ "licence" ] }, { "id": "mlcommons-mlperf-tiny-ds-cnn-kws", "name": "MLPerf Tiny DS-CNN (KWS)", "level": "model", "family_id": null, "developer": [ "mlcommons" ], "model_class": "TinyML", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [ "microcontroller" ], "footprint_note": "MCU-class", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MLPerf Tiny reference model for keyword spotting on microcontrollers and NPUs.", "verified_on": "2026-10-10", "sources": [ "https://mlcommons.org/benchmarks/inference-tiny/" ], "unverified": [ "licence" ] }, { "id": "mlcommons-mlperf-tiny-mobilenet-vww", "name": "MLPerf Tiny MobileNet (VWW)", "level": "model", "family_id": null, "developer": [ "mlcommons" ], "model_class": "TinyML", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [ "microcontroller" ], "footprint_note": "MCU-class", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MLPerf Tiny reference model for visual wake words on microcontrollers and NPUs.", "verified_on": "2026-10-10", "sources": [ "https://mlcommons.org/benchmarks/inference-tiny/" ], "unverified": [ "licence" ] }, { "id": "mlcommons-mlperf-tiny-resnet", "name": "MLPerf Tiny ResNet (CIFAR-10)", "level": "model", "family_id": null, "developer": [ "mlcommons" ], "model_class": "TinyML", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [ "microcontroller" ], "footprint_note": "MCU-class", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MLPerf Tiny reference model for image classification on microcontrollers and NPUs.", "verified_on": "2026-10-10", "sources": [ "https://mlcommons.org/benchmarks/inference-tiny/" ], "unverified": [ "licence" ] }, { "id": "mobilenet-v2", "name": "MobileNet v2", "level": "model", "family_id": null, "developer": [], "model_class": "TinyML", "modalities_in": [ "text" ], "modalities_out": [ "labels and boxes" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "microcontroller", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Model listed in Arm ML-Zoo, optimised for Arm IP (Cortex-A, Cortex-M, Mali, Ethos-U) in TensorFlow Lite format.", "verified_on": "2026-10-10", "sources": [ "https://github.com/ARM-software/ML-zoo" ], "unverified": [] }, { "id": "modernbert", "name": "ModernBERT", "level": "model", "family_id": null, "developer": [ "answer-ai", "lighton" ], "model_class": "embedding", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Encoder on which SEA-LION-ModernBERT is based; developer not stated in these sections.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/aisingapore", "https://arxiv.org/html/2412.13663" ], "unverified": [] }, { "id": "moondream-1-8b", "name": "Moondream 1.8B", "level": "model", "family_id": null, "developer": [], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "1.8B", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "CPU server", "laptop", "mobile or NPU" ], "footprint_note": "Ollama GGUF build", "runtimes": [ "ollama" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small model listed in the Ollama library for CPU, laptop and phone use.", "verified_on": "2026-10-10", "sources": [ "https://ollama.com/library" ], "unverified": [] }, { "id": "moonshot-ai-kimi", "name": "Kimi", "level": "family", "family_id": null, "developer": [ "moonshot-ai" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Kimi_(chatbot)", "https://huggingface.co/moonshotai" ], "unverified": [] }, { "id": "moonshot-ai-kimi-audio", "name": "Kimi-Audio", "level": "model", "family_id": null, "developer": [ "moonshot-ai" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Moonshot audio model, named but not verified further.", "verified_on": "2026-10-10", "sources": [ "https://select.keenable.ai/r/speech-recognition-accuracy-leaders-2022-2026/JbURQjlVooLQ6J10NkbRVw" ], "unverified": [] }, { "id": "moonshot-ai-kimi-k2", "name": "Kimi K2", "level": "model", "family_id": "moonshot-ai-kimi", "developer": [ "moonshot-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 1000000000000, "params_active": 32000000000, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-07", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Kimi_(chatbot)", "https://huggingface.co/moonshotai", "https://huggingface.co/moonshotai", "https://leaderboard.steel.dev/leaderboards/swe-bench-verified.md", "https://rits.shanghai.nyu.edu/ai/kimi-k3-open-weights-ship-2-8t-parameters-1-4-tb-to-run/" ], "unverified": [] }, { "id": "moonshot-ai-kimi-k2-0905", "name": "Kimi K2-0905", "level": "model", "family_id": "moonshot-ai-kimi", "developer": [ "moonshot-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Kimi_(chatbot)", "https://huggingface.co/moonshotai", "https://huggingface.co/moonshotai", "https://leaderboard.steel.dev/leaderboards/swe-bench-verified.md", "https://rits.shanghai.nyu.edu/ai/kimi-k3-open-weights-ship-2-8t-parameters-1-4-tb-to-run/" ], "unverified": [] }, { "id": "moonshot-ai-kimi-k2-5", "name": "Kimi K2.5", "level": "model", "family_id": "moonshot-ai-kimi", "developer": [ "moonshot-ai" ], "model_class": "multimodal LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 1000000000000, "params_active": null, "params_checkpoint": null, "params_note": "HF dates listed as May 19 / Apr 30 for K2.6 / K2.5 (year not stated; HF update dates)", "context_length": null, "release_date": "2026-01", "release_date_basis": "secondary report", "weights_date": "2026-04-30", "openness": "open weights restricted", "licence": "kimi-modified-mit", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Kimi_(chatbot)", "https://huggingface.co/moonshotai", "https://huggingface.co/moonshotai", "https://leaderboard.steel.dev/leaderboards/swe-bench-verified.md", "https://rits.shanghai.nyu.edu/ai/kimi-k3-open-weights-ship-2-8t-parameters-1-4-tb-to-run/" ], "unverified": [] }, { "id": "moonshot-ai-kimi-k2-6", "name": "Kimi K2.6", "level": "model", "family_id": "moonshot-ai-kimi", "developer": [ "moonshot-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 1000000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04", "release_date_basis": "secondary report", "weights_date": "2026-05-19", "openness": "open weights restricted", "licence": "kimi-modified-mit", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Kimi_(chatbot)", "https://huggingface.co/moonshotai", "https://huggingface.co/moonshotai", "https://leaderboard.steel.dev/leaderboards/swe-bench-verified.md", "https://rits.shanghai.nyu.edu/ai/kimi-k3-open-weights-ship-2-8t-parameters-1-4-tb-to-run/", "https://arena.ai/leaderboard/chat/text" ], "unverified": [] }, { "id": "moonshot-ai-kimi-k2-7-code", "name": "Kimi K2.7 Code", "level": "model", "family_id": "moonshot-ai-kimi", "developer": [ "moonshot-ai" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 1000000000000, "params_active": 32000000000, "params_checkpoint": null, "params_note": null, "context_length": 256000, "release_date": "2026-06", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights restricted", "licence": "kimi-modified-mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "multi-GPU node", "hyperscale cluster" ], "footprint_note": "T3/T2 estimate (unverified). HF updated Jun 15. Retired from Copilot 2026-10-02.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Coding fine-tune of Kimi K2.6.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/moonshotai/Kimi-K2.7-Code", "https://huggingface.co/moonshotai", "https://en.wikipedia.org/wiki/Kimi_(chatbot)", "https://docs.github.com/en/copilot/reference/ai-models/supported-models" ], "unverified": [ "deployment_tiers" ] }, { "id": "moonshot-ai-kimi-k2-thinking", "name": "Kimi K2 Thinking", "level": "model", "family_id": "moonshot-ai-kimi", "developer": [ "moonshot-ai" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Trained on H800 with InfiniBand; a circulating $4.6M cost was called unofficial by co-founder Yang Zhilin.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Moonshot AI reasoning model; its circulating training cost figure is unconfirmed by Moonshot.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Kimi_(chatbot)", "https://huggingface.co/moonshotai", "https://leaderboard.steel.dev/leaderboards/swe-bench-verified.md", "https://rits.shanghai.nyu.edu/ai/kimi-k3-open-weights-ship-2-8t-parameters-1-4-tb-to-run/", "https://www-web.itiger.com/news/1119147261", "https://huggingface.co/moonshotai/Kimi-K2-Thinking" ], "unverified": [ "params_note" ] }, { "id": "moonshot-ai-kimi-k3", "name": "Kimi K3", "level": "model", "family_id": "moonshot-ai-kimi", "developer": [ "moonshot-ai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "MoE, 896 experts, 16+2 active; KDA plus gated MLA hybrid attention", "params_total": 2800000000000, "params_active": 104000000000, "params_checkpoint": null, "params_note": null, "context_length": 1048576, "release_date": "2026-07-16", "release_date_basis": "inferred from page", "weights_date": "2026-07-27", "openness": "open weights restricted", "licence": "kimi-k3-license", "access_routes": [ "API", "open weights download" ], "deployment_tiers": [ "hyperscale cluster", "multi-GPU node" ], "footprint_note": "1.4 TB resident, 64+ accelerators (T2 multi-node cluster).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Moonshot AI's open-weight 2.8T MoE flagship; base for Cognition SWE-2 and available in GitHub Copilot. Video input on the card is inconsistent (register C4, not checked).", "verified_on": "2026-10-10", "sources": [ "https://www.kimi.com/blog/kimi-k3", "https://huggingface.co/api/models/moonshotai/Kimi-K3/commits/main", "https://letsdatascience.com/news/kimi-k3-debuts-as-28-trillion-parameter-open-weight-model-cac7da05", "https://artificialanalysis.ai/models/kimi-k3", "https://platform.kimi.ai/docs/pricing/chat", "https://rits.shanghai.nyu.edu/ai/kimi-k3-open-weights-ship-2-8t-parameters-1-4-tb-to-run/", "https://huggingface.co/moonshotai/Kimi-K3", "https://arena.ai/leaderboard/chat/text", "https://docs.github.com/en/copilot/reference/ai-models/supported-models", "https://huggingface.co/moonshotai", "https://huggingface.co/moonshotai/Kimi-K3/blob/main/LICENSE" ], "unverified": [] }, { "id": "morgan-stanley-lag-llama", "name": "Lag-Llama", "level": "model", "family_id": null, "developer": [ "morgan-stanley", "servicenow", "universite-de-montreal", "mila" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "dense transformer", "architecture_detail": "Decoder-only transformer using lags as covariates; univariate probabilistic", "params_total": 2450000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-10-12", "release_date_basis": "announcement page", "weights_date": "2024-02-07", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Early time-series foundation model (USA / Canada affiliations); worse than Seasonal Naive on GIFT-Eval MASE with test leakage flagged.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2310.08278v3", "https://huggingface.co/spaces/Salesforce/GIFT-Eval", "https://huggingface.co/api/models/time-series-foundation-models/Lag-Llama" ], "unverified": [] }, { "id": "moritzlaurer-mdeberta-v3-base-xnli-multilingual-nli-2mil7", "name": "mDeBERTa-v3-base-xnli-multilingual-nli-2mil7", "level": "model", "family_id": null, "developer": [ "moritzlaurer" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "mDeBERTa-v3-base encoder fine-tuned for natural language inference", "params_total": 278812163, "params_active": null, "params_checkpoint": null, "params_note": "Hugging Face API Safetensors count", "context_length": null, "release_date": "2022-08-22", "release_date_basis": "registry date", "weights_date": "2022-08-22", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "NLI on 100 languages; fine-tuned on 27 languages including Indonesian and Vietnamese", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "A multilingual natural language inference classifier for zero-shot classification, fine-tuned from mDeBERTa-v3-base on XNLI and 2.7 million multilingual NLI pairs.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/MoritzLaurer/mDeBERTa-v3-base-xnli-multilingual-nli-2mil7", "https://huggingface.co/api/models/MoritzLaurer/mDeBERTa-v3-base-xnli-multilingual-nli-2mil7" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "motif-technologies-motif-3", "name": "Motif-3", "level": "model", "family_id": null, "developer": [ "motif-technologies" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "Decoder-only MoE, 53 layers, Grouped Differential Latent Attention, 384 routed experts (top 8) plus 1 shared, MTP head", "params_total": 314000000000, "params_active": 13200000000, "params_checkpoint": null, "params_note": null, "context_length": 262144, "release_date": "2026-08", "release_date_basis": "inferred from page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English and Korean language tags", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Motif Technologies 314B-total, 13.2B-active MoE model post-trained for agentic, coding and reasoning tasks, released with open weights.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/Motif-Technologies/Motif-3" ], "unverified": [] }, { "id": "motif-technologies-motif-3-base", "name": "Motif-3-Base", "level": "model", "family_id": null, "developer": [ "motif-technologies" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Base checkpoint from which Motif-3 is post-trained.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/Motif-Technologies/Motif-3" ], "unverified": [] }, { "id": "mscred", "name": "MSCRED", "level": "model", "family_id": null, "developer": [ "university-of-notre-dame", "nec" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-11-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Top of the SKAB changepoint leaderboard.", "verified_on": "2026-10-10", "sources": [ "https://github.com/waico/SKAB", "https://arxiv.org/html/1811.08055", "https://arxiv.org/abs/1811.08055" ], "unverified": [] }, { "id": "mset", "name": "MSET (Multivariate State Estimation Technique)", "level": "model", "family_id": null, "developer": [], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Similarity-based anomaly detection method; widely associated with GE SmartSignal (unverified).", "verified_on": "2026-10-10", "sources": [ "https://github.com/waico/SKAB" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "multimedia-university-tf-resnet50-ganoderma", "name": "TF-ResNet50 Ganoderma detector", "level": "model", "family_id": null, "developer": [ "multimedia-university", "fgv-rd" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": "Fine-tuned ResNet50 after PCA on UAV hyperspectral imagery; SVM baseline", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "ResNet50 about 25M params (unverified)", "context_length": null, "release_date": "2025-10", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "research only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Research model for Ganoderma basal stem rot detection in oil palm (TENCON 2025): 84% accuracy, 74% sensitivity for early infection, 86% specificity.", "verified_on": "2026-10-10", "sources": [ "https://shdl.mmu.edu.my/15767/" ], "unverified": [] }, { "id": "mureka-v9", "name": "Mureka V9", "level": "model", "family_id": null, "developer": [], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Music model on AA Music Arena v1.1 Instrumental; developer unverified.", "verified_on": "2026-10-10", "sources": [ "https://alphasignal.ai/news/artificial-analysis-music-arena-v1-1-crowns-suno-v6-across-1-000-harder-prompts" ], "unverified": [ "developer" ] }, { "id": "mureka-v9-5", "name": "Mureka V9.5", "level": "model", "family_id": null, "developer": [], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Music model ranked 3rd on AA Music Arena v1.1 Vocal; developer unverified.", "verified_on": "2026-10-10", "sources": [ "https://alphasignal.ai/news/artificial-analysis-music-arena-v1-1-crowns-suno-v6-across-1-000-harder-prompts" ], "unverified": [ "developer" ] }, { "id": "nari-dia-1-6b", "name": "Nari Dia-1.6B", "level": "model", "family_id": null, "developer": [ "nari-labs" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open TTS model trending on Hugging Face (2025 to 2026).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/models?pipeline_tag=text-to-speech&sort=trending", "https://huggingface.co/nari-labs/Dia-1.6B" ], "unverified": [] }, { "id": "naver-hyperclova-x-seed-4b", "name": "HyperCLOVA X SEED 4B", "level": "model", "family_id": null, "developer": [ "naver" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 4000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-06-23", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://clova.ai/en/tech-blog" ], "unverified": [] }, { "id": "naver-hyperclova-x-seed-8b-omni", "name": "HyperCLOVA X SEED 8B Omni", "level": "model", "family_id": null, "developer": [ "naver" ], "model_class": "multimodal LLM", "modalities_in": [], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://clova.ai/en/tech-blog" ], "unverified": [] }, { "id": "naver-hyperclova-x-seed-think-32b", "name": "HyperCLOVA X SEED Think 32B", "level": "model", "family_id": null, "developer": [ "naver" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 32000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 128000, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://clova.ai/en/tech-blog", "https://artificialanalysis.ai/leaderboards/models" ], "unverified": [] }, { "id": "nec-siat", "name": "System Invariant Analytics Technology (SIAT)", "level": "model", "family_id": null, "developer": [ "nec" ], "model_class": "time series", "modalities_in": [ "time series", "sensor" ], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "NEC predictive maintenance and real-time equipment monitoring technology; proof of concept at Kalbe Farma (Apr to Jun 2026) met 2 of 3 targets per Kemenperin.", "verified_on": "2026-10-10", "sources": [ "https://suarapemerintah.id/2026/07/industri-farmasi-masuk-era-ai-kemenperin-fasilitasi-uji-coba-solusi-industri-4-0/" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "neuralgcm", "name": "NeuralGCM (0.7 and ENS)", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-11-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Weather model on WeatherBench 2 (2020).", "verified_on": "2026-10-10", "sources": [ "https://researchsites.withgoogle.com/weatherbench/deterministic/", "https://arxiv.org/html/2311.07222", "https://arxiv.org/abs/2311.07222" ], "unverified": [] }, { "id": "nih-nlm-trialgpt", "name": "TrialGPT", "level": "model", "family_id": null, "developer": [ "nih-nlm", "nci" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "research only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "LLM-based patient-to-trial matching from NIH; close to clinician accuracy on more than 1,000 patient-criterion pairs and 40% less screening time in a pilot (Nature Communications 2024). The base LLM is not named.", "verified_on": "2026-10-10", "sources": [ "https://www.nih.gov/news-events/news-releases/nih-developed-ai-algorithm-matches-potential-volunteers-clinical-trials" ], "unverified": [] }, { "id": "nii-llm-jp-llm-jp-4-1", "name": "LLM-jp-4.1", "level": "family", "family_id": null, "developer": [ "nii-llm-jp" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-10-04", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NII LLM-jp's 4.1 dense and MoE models; LLM-jp historically releases its corpora too (not checked for 4.1).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/llm-jp" ], "unverified": [ "licence" ] }, { "id": "nii-llm-jp-llm-jp-4-1-32b-a3b", "name": "LLM-jp-4.1 32B-A3B", "level": "model", "family_id": "nii-llm-jp-llm-jp-4-1", "developer": [ "nii-llm-jp" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 32000000000, "params_active": 3000000000, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-10-04", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "consumer GPU" ], "footprint_note": "GGUF quants", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "LLM-jp-4.1 32B-A3B, part of NII LLM-jp's 4.1 release.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/llm-jp" ], "unverified": [ "licence" ] }, { "id": "nii-llm-jp-llm-jp-4-1-33b", "name": "LLM-jp-4.1 33B", "level": "model", "family_id": "nii-llm-jp-llm-jp-4-1", "developer": [ "nii-llm-jp" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 33000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-10-04", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "consumer GPU" ], "footprint_note": "GGUF quants", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "LLM-jp-4.1 33B, part of NII LLM-jp's 4.1 release.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/llm-jp" ], "unverified": [ "licence" ] }, { "id": "nii-llm-jp-llm-jp-4-1-8b-thinking", "name": "LLM-jp-4.1 8B thinking", "level": "model", "family_id": "nii-llm-jp-llm-jp-4-1", "developer": [ "nii-llm-jp" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-10-04", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "consumer GPU" ], "footprint_note": "GGUF quants", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "LLM-jp-4.1 8B thinking, part of NII LLM-jp's 4.1 release.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/llm-jp" ], "unverified": [ "licence" ] }, { "id": "nii-llm-jp-llm-jp-4-vl-9b", "name": "LLM-jp-4-VL-9B", "level": "model", "family_id": null, "developer": [ "nii-llm-jp" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 9000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/llm-jp" ], "unverified": [] }, { "id": "nixtla-statsforecast", "name": "Nixtla statsforecast", "level": "model", "family_id": null, "developer": [ "nixtla" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": "ARIMA, ETS, Theta and other classical models", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Classical forecasting baseline library; production use OK.", "verified_on": "2026-10-10", "sources": [ "https://www.beri.net/article/timesfm-vs-chronos-vs-moirai-time-series-foundation-model-licence-bigquery-sagemaker-cost-2026" ], "unverified": [] }, { "id": "nixtla-timegpt", "name": "TimeGPT", "level": "family", "family_id": null, "developer": [ "nixtla" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "Client SDK Apache-2.0 (nixtla 0.9.0, 21 Sep 2026); README mentions self-deployment on own infrastructure (terms unverified)", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Nixtla proprietary forecasting model via API, now TimeGPT-2.1; no Nixtla submission on GIFT-Eval.", "verified_on": "2026-10-10", "sources": [ "https://www.nixtla.io/docs", "https://arxiv.org/abs/2310.03589", "https://github.com/Nixtla/nixtla", "https://pypi.org/pypi/nixtla/json" ], "unverified": [] }, { "id": "nixtla-timegpt-1", "name": "TimeGPT-1", "level": "model", "family_id": "nixtla-timegpt", "developer": [ "nixtla" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-10-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First TimeGPT.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2310.03589" ], "unverified": [] }, { "id": "nixtla-timegpt-2", "name": "TimeGPT-2", "level": "model", "family_id": "nixtla-timegpt", "developer": [ "nixtla" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "TimeGPT successor.", "verified_on": "2026-10-10", "sources": [ "https://github.com/Nixtla/nixtla" ], "unverified": [] }, { "id": "nixtla-timegpt-2-1", "name": "TimeGPT-2.1", "level": "model", "family_id": "nixtla-timegpt", "developer": [ "nixtla" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Current TimeGPT version.", "verified_on": "2026-10-10", "sources": [ "https://github.com/Nixtla/nixtla" ], "unverified": [] }, { "id": "nomic-ai-nomic-embed-multimodal", "name": "nomic-embed-multimodal", "level": "model", "family_id": null, "developer": [ "nomic-ai" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "3B and 7B", "context_length": null, "release_date": "2025", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Nomic AI's multimodal embedding models (3B and 7B).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/nomic-ai" ], "unverified": [ "licence" ] }, { "id": "nomic-ai-nomic-embed-text-v2-moe", "name": "Nomic embed text v2 MoE", "level": "model", "family_id": null, "developer": [ "nomic-ai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 500000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Nomic AI's MoE text embedding model; no 2026 release found.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/nomic-ai" ], "unverified": [ "licence" ] }, { "id": "ntt-tsuzumi", "name": "tsuzumi", "level": "model", "family_id": null, "developer": [ "ntt" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Not researched (search budget).", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/leaderboards/models" ], "unverified": [ "status" ] }, { "id": "numenta-htm", "name": "Numenta HTM", "level": "model", "family_id": null, "developer": [], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Numenta's HTM detector on NAB.", "verified_on": "2026-10-10", "sources": [ "https://github.com/numenta/NAB" ], "unverified": [] }, { "id": "nvidia-alpamayo-1-5-10b", "name": "Alpamayo 1.5-10B", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "driving", "modalities_in": [ "video", "text", "sensor" ], "modalities_out": [ "action", "text" ], "architecture": "hybrid", "architecture_detail": "Transformer VLA with diffusion trajectory decoder; 4 cameras, egomotion, text inputs; Chain-of-Causation reasoning plus 6.4 s, 64-waypoint trajectory", "params_total": 10000000000, "params_active": null, "params_checkpoint": null, "params_note": "About 10B (8.2B backbone + 2.3B action expert)", "context_length": null, "release_date": "2026-03-19", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": "openmdw-1-1", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [ "nvidia-nim" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA open driving VLA on a Cosmos-Reason2 backbone, RL post-trained; weights OpenMDW-1.1, code Apache 2.0.", "verified_on": "2026-10-10", "sources": [ "https://catalog.ngc.nvidia.com/orgs/nim/nvidia/models/alpamayo-1-5-10b/-" ], "unverified": [] }, { "id": "nvidia-alpamayo-r1", "name": "Alpamayo-R1", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "driving", "modalities_in": [ "video", "text" ], "modalities_out": [ "action", "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-12", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA open driving VLA with reasoning.", "verified_on": "2026-10-10", "sources": [ "https://www.opensourceforu.com/2025/12/nvidia-drops-alpamayo-r1-as-open-source-leap-toward-level-4-autonomy/" ], "unverified": [ "openness", "licence" ] }, { "id": "nvidia-canary-1b", "name": "Canary-1B", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "FastConformer AED", "params_total": 1000000000, "params_active": null, "params_checkpoint": null, "params_note": "about 1B", "context_length": null, "release_date": "2024", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA ASR model that led the Open ASR Leaderboard in 2024.", "verified_on": "2026-10-10", "sources": [ "https://benchmarklist.com/benchmarks/open_asr_leaderboard/", "https://select.keenable.ai/r/speech-recognition-accuracy-leaders-2022-2026/JbURQjlVooLQ6J10NkbRVw" ], "unverified": [ "licence" ] }, { "id": "nvidia-canary-1b-flash", "name": "Canary-1B-flash", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "FastConformer AED", "params_total": 811000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03-07", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA FastConformer attention encoder-decoder ASR model.", "verified_on": "2026-10-10", "sources": [ "https://benchmarklist.com/benchmarks/open_asr_leaderboard/", "https://huggingface.co/api/models/nvidia/canary-1b-flash" ], "unverified": [] }, { "id": "nvidia-canary-qwen-2-5b", "name": "Canary-Qwen-2.5B", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "autoregressive", "architecture_detail": "FastConformer encoder + Qwen LLM decoder (unverified detail)", "params_total": 2500000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-06", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA ASR model with Canary encoder and Qwen decoder; led the Open ASR Leaderboard in 2025.", "verified_on": "2026-10-10", "sources": [ "https://www.gladia.io/blog/best-open-source-speech-to-text-models", "https://benchmarklist.com/benchmarks/open_asr_leaderboard/", "https://select.keenable.ai/r/speech-recognition-accuracy-leaders-2022-2026/JbURQjlVooLQ6J10NkbRVw" ], "unverified": [] }, { "id": "nvidia-colembed-v2", "name": "ColEmbed V2", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 1000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA ColEmbed V2 on a Llama base; NVIDIA licence unverified.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/nvidia", "https://huggingface.co/nvidia/models?search=nemotron-3" ], "unverified": [ "licence" ] }, { "id": "nvidia-cosmos", "name": "Cosmos", "level": "family", "family_id": null, "developer": [ "nvidia" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-01", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA Cosmos world foundation models for physical AI.", "verified_on": "2026-10-10", "sources": [ "https://investor.nvidia.com/news/press-release-details/2026/NVIDIA-Launches-Cosmos-3-the-Open-Frontier-Foundation-Model-for-Physical-AI/default.aspx", "https://forums.developer.nvidia.com/t/nvidia-cosmos-announcements-at-ces-2026/356629", "https://arxiv.org/abs/2501.03575" ], "unverified": [] }, { "id": "nvidia-cosmos-3", "name": "Cosmos 3", "level": "family", "family_id": "nvidia-cosmos", "developer": [ "nvidia" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-06-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA Cosmos 3 (Super, Nano; Edge coming soon).", "verified_on": "2026-10-10", "sources": [ "https://investor.nvidia.com/news/press-release-details/2026/NVIDIA-Launches-Cosmos-3-the-Open-Frontier-Foundation-Model-for-Physical-AI/default.aspx" ], "unverified": [] }, { "id": "nvidia-cosmos-3-edge", "name": "Cosmos 3 Edge", "level": "model", "family_id": "nvidia-cosmos-3", "developer": [ "nvidia" ], "model_class": "world model", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "video", "action", "text" ], "architecture": "hybrid", "architecture_detail": "Unifies reasoning VLM, world generation and action prediction (world action model backbone)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "announced", "summary": "NVIDIA open world foundation model for physical AI, on Hugging Face, build.nvidia.com and NIM; NVIDIA claims #1 among open models on several physics and robotics benchmarks without scores.", "verified_on": "2026-10-10", "sources": [ "https://investor.nvidia.com/news/press-release-details/2026/NVIDIA-Launches-Cosmos-3-the-Open-Frontier-Foundation-Model-for-Physical-AI/default.aspx" ], "unverified": [] }, { "id": "nvidia-cosmos-3-nano", "name": "Cosmos 3 Nano", "level": "model", "family_id": "nvidia-cosmos-3", "developer": [ "nvidia" ], "model_class": "world model", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "video", "action", "text" ], "architecture": "hybrid", "architecture_detail": "Unifies reasoning VLM, world generation and action prediction (world action model backbone)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-06-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "API", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [ "nvidia-nim" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA open world foundation model for physical AI, on Hugging Face, build.nvidia.com and NIM; NVIDIA claims #1 among open models on several physics and robotics benchmarks without scores.", "verified_on": "2026-10-10", "sources": [ "https://investor.nvidia.com/news/press-release-details/2026/NVIDIA-Launches-Cosmos-3-the-Open-Frontier-Foundation-Model-for-Physical-AI/default.aspx" ], "unverified": [] }, { "id": "nvidia-cosmos-3-super", "name": "Cosmos 3 Super", "level": "model", "family_id": "nvidia-cosmos-3", "developer": [ "nvidia" ], "model_class": "world model", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "video", "action", "text" ], "architecture": "hybrid", "architecture_detail": "Unifies reasoning VLM, world generation and action prediction (world action model backbone)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-06-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "API", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [ "nvidia-nim" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA open world foundation model for physical AI, on Hugging Face, build.nvidia.com and NIM; NVIDIA claims #1 among open models on several physics and robotics benchmarks without scores.", "verified_on": "2026-10-10", "sources": [ "https://investor.nvidia.com/news/press-release-details/2026/NVIDIA-Launches-Cosmos-3-the-Open-Frontier-Foundation-Model-for-Physical-AI/default.aspx" ], "unverified": [] }, { "id": "nvidia-cosmos-predict-1", "name": "Cosmos Predict 1", "level": "model", "family_id": "nvidia-cosmos", "developer": [ "nvidia" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First Cosmos Predict model.", "verified_on": "2026-10-10", "sources": [ "https://forums.developer.nvidia.com/t/nvidia-cosmos-announcements-at-ces-2026/356629" ], "unverified": [] }, { "id": "nvidia-cosmos-predict-2", "name": "Cosmos Predict 2", "level": "model", "family_id": "nvidia-cosmos", "developer": [ "nvidia" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier Cosmos Predict model.", "verified_on": "2026-10-10", "sources": [ "https://forums.developer.nvidia.com/t/nvidia-cosmos-announcements-at-ces-2026/356629" ], "unverified": [] }, { "id": "nvidia-cosmos-predict-2-5", "name": "Cosmos Predict 2.5", "level": "model", "family_id": "nvidia-cosmos", "developer": [ "nvidia" ], "model_class": "world model", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA unified text-, image- and video-to-world model trained on 200M clips; licence name not stated.", "verified_on": "2026-10-10", "sources": [ "https://forums.developer.nvidia.com/t/nvidia-cosmos-announcements-at-ces-2026/356629", "https://github.com/nvidia-cosmos/cosmos-transfer2.5" ], "unverified": [] }, { "id": "nvidia-cosmos-reason-1", "name": "Cosmos Reason 1", "level": "model", "family_id": "nvidia-cosmos", "developer": [ "nvidia" ], "model_class": "multimodal LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier Cosmos Reason model.", "verified_on": "2026-10-10", "sources": [ "https://forums.developer.nvidia.com/t/nvidia-cosmos-announcements-at-ces-2026/356629" ], "unverified": [] }, { "id": "nvidia-cosmos-reason-2", "name": "Cosmos Reason 2", "level": "model", "family_id": "nvidia-cosmos", "developer": [ "nvidia" ], "model_class": "multimodal LLM", "modalities_in": [ "image", "video", "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "2B and 8B", "context_length": null, "release_date": "2026-01-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA physical-AI reasoning VLM; backbone of Alpamayo 1.5.", "verified_on": "2026-10-10", "sources": [ "https://forums.developer.nvidia.com/t/nvidia-cosmos-announcements-at-ces-2026/356629", "https://catalog.ngc.nvidia.com/orgs/nim/nvidia/models/alpamayo-1-5-10b/-" ], "unverified": [] }, { "id": "nvidia-cosmos-transfer-1", "name": "Cosmos Transfer 1", "level": "model", "family_id": "nvidia-cosmos", "developer": [ "nvidia" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier Cosmos Transfer model.", "verified_on": "2026-10-10", "sources": [ "https://forums.developer.nvidia.com/t/nvidia-cosmos-announcements-at-ces-2026/356629" ], "unverified": [] }, { "id": "nvidia-cosmos-transfer-2-5", "name": "Cosmos Transfer 2.5", "level": "model", "family_id": "nvidia-cosmos", "developer": [ "nvidia" ], "model_class": "world model", "modalities_in": [ "video" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": 2000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA world-to-world translation model (sim to photoreal).", "verified_on": "2026-10-10", "sources": [ "https://forums.developer.nvidia.com/t/nvidia-cosmos-announcements-at-ces-2026/356629", "https://github.com/nvidia-cosmos/cosmos-transfer2.5" ], "unverified": [] }, { "id": "nvidia-eagle-2-5", "name": "Eagle 2.5", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "multimodal LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA VLM cited in the GR00T N1.6 lineage.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/nvidia/GR00T-N1.6-3B" ], "unverified": [] }, { "id": "nvidia-gr00t", "name": "GR00T", "level": "family", "family_id": null, "developer": [ "nvidia" ], "model_class": "robot policy", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA humanoid robot foundation models.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2503.14734", "https://huggingface.co/nvidia/GR00T-N1.6-3B" ], "unverified": [] }, { "id": "nvidia-gr00t-n1", "name": "GR00T N1", "level": "model", "family_id": "nvidia-gr00t", "developer": [ "nvidia" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": "hybrid", "architecture_detail": "Dual-system VLA: VLM (System 2) plus diffusion transformer (System 1)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03-18", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA open humanoid foundation model deployed on Fourier GR-1.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2503.14734" ], "unverified": [] }, { "id": "nvidia-gr00t-n1-6", "name": "GR00T N1.6", "level": "model", "family_id": "nvidia-gr00t", "developer": [ "nvidia" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": "hybrid", "architecture_detail": "SigLIP2 vision, T5 text encoder, embodiment-indexed MLPs, flow-matching DiT action head", "params_total": 3000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "nvidia-gr00t-noncommercial", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA 3B open VLA citing Eagle 2.5 VLM and integrated with Cosmos Reason.", "verified_on": "2026-10-10", "sources": [ "https://forums.developer.nvidia.com/t/nvidia-cosmos-announcements-at-ces-2026/356629", "https://huggingface.co/nvidia/GR00T-N1.6-3B" ], "unverified": [] }, { "id": "nvidia-gr00t-n1-7", "name": "GR00T N1.7", "level": "model", "family_id": "nvidia-gr00t", "developer": [ "nvidia" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-03-21", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "gated preview" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [ "lerobot" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": "VLA for dexterous and full-body humanoid control in early access, with commercial licensing and LeRobot integration.", "verified_on": "2026-10-10", "sources": [ "https://www.originofbots.com/news/nvidia-gr00t-n17-hits-early-access-for-humanoids" ], "unverified": [] }, { "id": "nvidia-gr00t-n2", "name": "GR00T N2", "level": "model", "family_id": "nvidia-gr00t", "developer": [ "nvidia" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "announced", "summary": "Previewed successor with a more-than-2x claim, planned for end of 2026.", "verified_on": "2026-10-10", "sources": [ "https://www.originofbots.com/news/nvidia-gr00t-n17-hits-early-access-for-humanoids" ], "unverified": [] }, { "id": "nvidia-llama-3-1-nemotron-70b-instruct", "name": "Llama-3.1-Nemotron-70B-Instruct", "level": "model", "family_id": "nvidia-nemotron", "developer": [ "nvidia" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "70B per model name", "context_length": 128000, "release_date": "2024-10-12", "release_date_basis": "registry date", "weights_date": "2024-10-12", "openness": "open weights restricted", "licence": "nvidia-open-model-license", "access_routes": [ "open weights download" ], "deployment_tiers": [ "multi-GPU node" ], "footprint_note": "Needs 2 or more 80GB GPUs (Ampere or newer) and 150GB disk with Transformers; also offered as an NVIDIA NIM container", "runtimes": [ "nvidia-nim" ], "languages_note": "Hugging Face language tag: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA customization of Llama 3.1 70B Instruct trained with RLHF (REINFORCE) to improve the helpfulness of responses. Use is governed by the NVIDIA Open Model License with the Llama 3.1 Community License as additional terms.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/nvidia/Llama-3.1-Nemotron-70B-Instruct-HF", "https://catalog.ngc.nvidia.com/orgs/nim/teams/nvidia/containers/llama-3.1-nemotron-70b-instruct" ], "unverified": [] }, { "id": "nvidia-llama-3-1-nemotron-ultra-253b-v1", "name": "Llama 3.1 Nemotron Ultra 253B v1", "level": "model", "family_id": "nvidia-nemotron", "developer": [ "nvidia" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "Derived from Llama-3.1-405B-Instruct by Neural Architecture Search, knowledge distillation (65B tokens) and continual pretraining (88B tokens), then SFT and GRPO", "params_total": 253000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 131072, "release_date": "2025-04-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "nvidia-open-model-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English and coding languages; also German, French, Italian, Portuguese, Hindi, Spanish, Thai", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA 253B reasoning model derived from Llama 3.1 405B Instruct, ready for commercial use under the NVIDIA Open Model License plus the Llama 3.1 Community License.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/nvidia/Llama-3_1-Nemotron-Ultra-253B-v1" ], "unverified": [] }, { "id": "nvidia-llama-3-3-nemotron-super-49b-v1", "name": "Llama 3.3 Nemotron Super 49B v1", "level": "model", "family_id": "nvidia-nemotron", "developer": [ "nvidia" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "Derived from Llama-3.3-70B-Instruct by Neural Architecture Search and block-wise distillation, knowledge distillation on 40B tokens, SFT and RL", "params_total": 49000000000, "params_active": null, "params_checkpoint": null, "params_note": "Hugging Face lists 50B", "context_length": 131072, "release_date": "2025-03-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "nvidia-open-model-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English and coding languages; also German, French, Italian, Portuguese, Hindi, Spanish, Thai", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA 49B reasoning model derived from Llama 3.3 70B Instruct, under the NVIDIA Open Model License plus the Llama 3.3 Community License.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/nvidia/Llama-3_3-Nemotron-Super-49B-v1" ], "unverified": [] }, { "id": "nvidia-llama-nemotron-embed-vl-1b-v2", "name": "Llama Nemotron Embed VL 1B v2", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 1000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA Llama Nemotron Embed VL 1B v2 on a Llama base; NVIDIA licence unverified.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/nvidia", "https://huggingface.co/nvidia/models?search=nemotron-3" ], "unverified": [ "licence" ] }, { "id": "nvidia-llama-nemotron-rerank-vl-1b-v2", "name": "Llama Nemotron Rerank VL 1B v2", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "reranker", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 1000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA Llama Nemotron Rerank VL 1B v2 on a Llama base; NVIDIA licence unverified.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/nvidia", "https://huggingface.co/nvidia/models?search=nemotron-3" ], "unverified": [ "licence" ] }, { "id": "nvidia-lyra-2-0", "name": "Lyra-2.0", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "3D", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-07", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA 3D model trending on Hugging Face.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/models?pipeline_tag=image-to-3d&sort=trending", "https://en.wikipedia.org/wiki/Nvidia" ], "unverified": [] }, { "id": "nvidia-nemotron", "name": "Nemotron", "level": "family", "family_id": null, "developer": [ "nvidia" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "NVIDIA model family named for Siemens EDA software copilots.", "verified_on": "2026-10-10", "sources": [ "https://via.tt.se/pressmeddelande/4202605/siemens-unveils-technologies-to-accelerate-the-industrial-ai-revolution-at-ces-2026%3FpublisherId%3D259167%26lang%3Den" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "nvidia-nemotron-3", "name": "Nemotron 3", "level": "family", "family_id": null, "developer": [ "nvidia" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Nano, Super and Ultra sizes.", "context_length": null, "release_date": "2025-12-04", "release_date_basis": "registry date", "weights_date": null, "openness": "open source", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA's Nemotron 3 family of hybrid Mamba-Transformer MoE models, partially open on data.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/nvidia", "https://huggingface.co/nvidia/models?search=nemotron-3", "https://www.nvidia.com/en-us/ai-data-science/foundation-models/nemotron/", "https://huggingface.co/api/models/nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16", "https://huggingface.co/api/models/nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16", "https://huggingface.co/api/models/nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16", "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16", "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16" ], "unverified": [] }, { "id": "nvidia-nemotron-3-5-content-safety", "name": "Nemotron 3.5 Content Safety", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "guard", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 4000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-06-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "openmdw-1-1", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "12 languages; no Indonesian", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA's 4B multimodal content safety model with the Aegis 2.0 taxonomy plus a custom-policy mode (safety and prompt injection); licence OpenMDW 1.1 plus Gemma Terms of Use. HF 24 Jun 2026.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/nvidia/Nemotron-3.5-Content-Safety" ], "unverified": [] }, { "id": "nvidia-nemotron-3-5-lightning-30b-a3b", "name": "Nemotron 3.5 Lightning 30B-A3B", "level": "model", "family_id": "nvidia-nemotron-3", "developer": [ "nvidia" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Hybrid Mamba MoE", "params_total": 30000000000, "params_active": 3500000000, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026", "release_date_basis": "notes", "weights_date": null, "openness": "open source", "licence": "openmdw-1-1", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": "Consumer GPU at NVFP4 (listed at 18B)", "runtimes": [], "languages_note": null, "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Nemotron 3.5 Lightning hybrid Mamba MoE model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/nvidia", "https://huggingface.co/nvidia/models?search=nemotron-3", "https://leaderboard.sea-lion.ai/detailed/SEA", "https://leaderboard.sea-lion.ai/detailed/ID" ], "unverified": [ "licence", "indonesia_hosting" ], "indonesian_measured": [ { "benchmark": "SEA-HELM Indonesian leaderboard (ID aggregate, first score column of the Indonesian Competency and Task Scores table)", "score": "64.34 (95% CI -1.36 to +1.37)", "metric": "score 0 to 100, average of 2000 bootstraps with 95% CI", "run_by": "AI Singapore (SEA-HELM leaderboard maintainer)", "url": "https://leaderboard.sea-lion.ai/detailed/ID" } ] }, { "id": "nvidia-nemotron-3-content-safety-4b", "name": "Nemotron 3 Content Safety 4B", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "guard", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 4000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-03", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/nvidia", "https://huggingface.co/nvidia/models?search=nemotron-3" ], "unverified": [] }, { "id": "nvidia-nemotron-3-embed-1b", "name": "Nemotron-3-Embed-1B", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 1000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 32768, "release_date": "2026-07-16", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "openmdw-1-1", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "34 languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA Nemotron-3-Embed-1B text embedding model, 4096 dimensions (sliceable to 2048 or 1024); OpenMDW-1.1 (Apache 2.0 noted).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/nvidia/Nemotron-3-Embed-8B-BF16" ], "unverified": [] }, { "id": "nvidia-nemotron-3-embed-8b", "name": "Nemotron-3-Embed-8B", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 32768, "release_date": "2026-07-16", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "openmdw-1-1", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "34 languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA Nemotron-3-Embed-8B text embedding model, 4096 dimensions (sliceable to 2048 or 1024); OpenMDW-1.1 (Apache 2.0 noted).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/nvidia/Nemotron-3-Embed-8B-BF16" ], "unverified": [] }, { "id": "nvidia-nemotron-3-nano-30b-a3b", "name": "Nemotron 3 Nano 30B-A3B", "level": "model", "family_id": "nvidia-nemotron-3", "developer": [ "nvidia" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Hybrid Mamba MoE", "params_total": 30000000000, "params_active": 3000000000, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-12", "release_date_basis": "notes", "weights_date": null, "openness": "open source", "licence": "openmdw-1-1", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": "Consumer GPU at NVFP4 (listed at 18B)", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Base for Nemotron-SEA-LION v4.8-30B-A3B.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/nvidia", "https://huggingface.co/nvidia/models?search=nemotron-3" ], "unverified": [ "licence" ] }, { "id": "nvidia-nemotron-3-nano-4b", "name": "Nemotron 3 Nano 4B", "level": "model", "family_id": "nvidia-nemotron-3", "developer": [ "nvidia" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Hybrid Mamba MoE", "params_total": 4000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-03", "release_date_basis": "notes", "weights_date": null, "openness": "open source", "licence": "openmdw-1-1", "access_routes": [ "open weights download" ], "deployment_tiers": [ "laptop", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small Nemotron 3 model for CPU and edge.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/nvidia", "https://huggingface.co/nvidia/models?search=nemotron-3" ], "unverified": [ "licence" ] }, { "id": "nvidia-nemotron-3-nano-omni", "name": "Nemotron 3 Nano Omni", "level": "model", "family_id": "nvidia-nemotron-3", "developer": [ "nvidia" ], "model_class": "multimodal LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Hybrid Mamba MoE", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04", "release_date_basis": "notes", "weights_date": null, "openness": "open source", "licence": "openmdw-1-1", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": "Consumer GPU at NVFP4 (listed at 18B)", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multimodal Nemotron 3 Nano model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/nvidia", "https://huggingface.co/nvidia/models?search=nemotron-3" ], "unverified": [ "licence" ] }, { "id": "nvidia-nemotron-3-super", "name": "Nemotron 3 Super", "level": "model", "family_id": "nvidia-nemotron-3", "developer": [ "nvidia" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Hybrid Mamba-Transformer MoE", "params_total": 120000000000, "params_active": 12000000000, "params_checkpoint": null, "params_note": null, "context_length": 262000, "release_date": "2026", "release_date_basis": "registry date", "weights_date": null, "openness": "open source", "licence": "openmdw-1-1", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": "Single GPU at NVFP4 (listed at 67B)", "runtimes": [], "languages_note": null, "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "in-region Jakarta", "status": "active", "summary": "NVIDIA's 120B hybrid MoE model; base for Nemotron-SEA-LION v4.8.", "verified_on": "2026-10-10", "sources": [ "https://codersera.com/blog/open-source-llms-landscape-2026/", "https://huggingface.co/nvidia", "https://huggingface.co/nvidia/models?search=nemotron-3", "https://leaderboard.sea-lion.ai/detailed/SEA", "https://leaderboard.sea-lion.ai/detailed/ID", "https://docs.aws.amazon.com/bedrock/latest/userguide/models-region-compatibility.md" ], "unverified": [ "context_length" ], "hosting_provider": "Amazon Bedrock (ap-southeast-3 Jakarta)", "hosting_url": "https://docs.aws.amazon.com/bedrock/latest/userguide/models-region-compatibility.md", "indonesian_measured": [ { "benchmark": "SEA-HELM Indonesian leaderboard (ID aggregate, first score column of the Indonesian Competency and Task Scores table)", "score": "70.78 (95% CI -1.22 to +1.20)", "metric": "score 0 to 100, average of 2000 bootstraps with 95% CI", "run_by": "AI Singapore (SEA-HELM leaderboard maintainer)", "url": "https://leaderboard.sea-lion.ai/detailed/ID" } ] }, { "id": "nvidia-nemotron-3-ultra", "name": "Nemotron 3 Ultra 550B A55B", "level": "model", "family_id": "nvidia-nemotron-3", "developer": [ "nvidia" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "LatentMoE: Mamba-2 + MoE + attention hybrid, MTP", "params_total": 550000000000, "params_active": 55000000000, "params_checkpoint": null, "params_note": "From AA model name", "context_length": 1000000, "release_date": "2026-06-04", "release_date_basis": "secondary report", "weights_date": null, "openness": "open source", "licence": "openmdw-1-1", "access_routes": [ "open weights download" ], "deployment_tiers": [ "multi-GPU node" ], "footprint_note": "NVFP4 variant listed at 303B", "runtimes": [], "languages_note": "10 to 12 languages listed; no Indonesian", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA open-weights model (AA: open weights model, released June 2026).", "verified_on": "2026-10-10", "sources": [ "https://leaderboard.steel.dev/leaderboards/swe-bench-verified.md", "https://artificialanalysis.ai/models/nvidia-nemotron-3-ultra-550b-a55b", "https://artificialanalysis.ai/leaderboards/models", "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16", "https://huggingface.co/nvidia", "https://huggingface.co/nvidia/models?search=nemotron-3", "https://huggingface.co/api/models/nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16" ], "unverified": [] }, { "id": "nvidia-nemotron-3-ultra-base", "name": "Nemotron 3 Ultra Base", "level": "variant", "family_id": "nvidia-nemotron-3", "developer": [ "nvidia" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open source", "licence": "openmdw-1-1", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16", "https://huggingface.co/nvidia", "https://huggingface.co/nvidia/models?search=nemotron-3" ], "unverified": [] }, { "id": "nvidia-nemotron-3-ultra-genrm", "name": "Nemotron 3 Ultra GenRM", "level": "variant", "family_id": "nvidia-nemotron-3", "developer": [ "nvidia" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open source", "licence": "openmdw-1-1", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16", "https://huggingface.co/nvidia", "https://huggingface.co/nvidia/models?search=nemotron-3" ], "unverified": [] }, { "id": "nvidia-nemotron-3-ultra-nvfp4", "name": "Nemotron 3 Ultra NVFP4", "level": "variant", "family_id": "nvidia-nemotron-3", "developer": [ "nvidia" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open source", "licence": "openmdw-1-1", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16", "https://huggingface.co/nvidia", "https://huggingface.co/nvidia/models?search=nemotron-3" ], "unverified": [] }, { "id": "nvidia-nemotron-4-340b", "name": "Nemotron-4 340B", "level": "model", "family_id": "nvidia-nemotron", "developer": [ "nvidia" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "Decoder-only transformer with GQA and RoPE; family has Base, Instruct and Reward models", "params_total": 340000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 4096, "release_date": "2024-06-14", "release_date_basis": "announcement page", "weights_date": "2024-06-13", "openness": "open weights permissive", "licence": "nvidia-open-model-license", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [ "multi-GPU node" ], "footprint_note": "BF16 inference: 8x H200 (one node), 16x H100 or 16x A100 80GB (two nodes)", "runtimes": [], "languages_note": "Instruct model intended for English; base pre-trained on English text, 50+ natural languages and 40+ coding languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Nemotron-4 340B is an NVIDIA open model family (Base, Instruct, Reward) built to generate synthetic training data for other LLMs.", "verified_on": "2026-10-11", "sources": [ "https://blogs.nvidia.com/blog/nemotron-4-synthetic-data-generation-llm-training/", "https://huggingface.co/nvidia/Nemotron-4-340B-Instruct", "https://huggingface.co/api/models/nvidia/Nemotron-4-340B-Instruct" ], "unverified": [] }, { "id": "nvidia-nemotron-4-340b-base", "name": "Nemotron-4-340B-Base", "level": "model", "family_id": "nvidia-nemotron", "developer": [ "nvidia" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": 340000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "nvidia-open-model-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Base model of Nemotron-4-340B-Instruct, pre-trained on 9 trillion tokens.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/nvidia/Nemotron-4-340B-Instruct", "https://blogs.nvidia.com/blog/nemotron-4-synthetic-data-generation-llm-training/" ], "unverified": [] }, { "id": "nvidia-nemotron-retriever", "name": "Nemotron Retriever models", "level": "family", "family_id": null, "developer": [ "nvidia" ], "model_class": "embedding", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "NVIDIA retriever models shipped for RAG.", "verified_on": "2026-10-10", "sources": [ "https://www.nvidia.com/en-us/ai-data-science/foundation-models/nemotron/" ], "unverified": [] }, { "id": "nvidia-nemotron-safety", "name": "Nemotron Safety models", "level": "family", "family_id": null, "developer": [ "nvidia" ], "model_class": "guard", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "NVIDIA safety classifier models listed on the Nemotron page.", "verified_on": "2026-10-10", "sources": [ "https://www.nvidia.com/en-us/ai-data-science/foundation-models/nemotron/" ], "unverified": [] }, { "id": "nvidia-parakeet-tdt-0-6b", "name": "Parakeet-TDT-0.6B", "level": "family", "family_id": null, "developer": [ "nvidia" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "FastConformer + TDT (CTC/TDT), much faster than LLM-decoder models", "params_total": 600000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU", "CPU server" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA fast ASR model; v3 adds 25 European languages.", "verified_on": "2026-10-10", "sources": [ "https://www.gladia.io/blog/best-open-source-speech-to-text-models", "https://benchmarklist.com/benchmarks/open_asr_leaderboard/" ], "unverified": [] }, { "id": "nvidia-parakeet-tdt-0-6b-v2", "name": "Parakeet-TDT-0.6B v2", "level": "variant", "family_id": "nvidia-parakeet-tdt-0-6b", "developer": [ "nvidia" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": 600000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU", "CPU server" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Parakeet-TDT-0.6B v2.", "verified_on": "2026-10-10", "sources": [ "https://www.gladia.io/blog/best-open-source-speech-to-text-models", "https://benchmarklist.com/benchmarks/open_asr_leaderboard/" ], "unverified": [] }, { "id": "nvidia-parakeet-tdt-0-6b-v3", "name": "Parakeet-TDT-0.6B v3", "level": "variant", "family_id": "nvidia-parakeet-tdt-0-6b", "developer": [ "nvidia" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": 600000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU", "CPU server" ], "footprint_note": null, "runtimes": [], "languages_note": "adds 25 European languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Parakeet-TDT-0.6B v3.", "verified_on": "2026-10-10", "sources": [ "https://www.gladia.io/blog/best-open-source-speech-to-text-models", "https://benchmarklist.com/benchmarks/open_asr_leaderboard/" ], "unverified": [] }, { "id": "nvidia-peoplenet", "name": "PeopleNet", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA TAO model in the NGC catalogue.", "verified_on": "2026-10-10", "sources": [ "https://catalog.ngc.nvidia.com/models" ], "unverified": [] }, { "id": "nvidia-sortformer", "name": "Sortformer", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "speaker", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA NeMo open speaker diarization model; exact licence unverified.", "verified_on": "2026-10-10", "sources": [ "https://www.pyannote.ai/blog/top-7-speaker-diarization-apis-and-models-in-2026" ], "unverified": [ "licence" ] }, { "id": "nvidia-trafficcamnet", "name": "TrafficCamNet", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA TAO model in the NGC catalogue.", "verified_on": "2026-10-10", "sources": [ "https://catalog.ngc.nvidia.com/models" ], "unverified": [] }, { "id": "nx-ai-tirex", "name": "TiRex", "level": "model", "family_id": null, "developer": [ "nx-ai" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": "xLSTM-based", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-05-26", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "nx-ai-community-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NX-AI xLSTM time-series model (Linz; Chief Scientist Sepp Hochreiter).", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2605.20119", "https://www.nx-ai.com/", "https://huggingface.co/spaces/Salesforce/GIFT-Eval", "https://huggingface.co/api/models/NX-AI/TiRex" ], "unverified": [] }, { "id": "nx-ai-tirex-2", "name": "TiRex-2", "level": "model", "family_id": null, "developer": [ "nx-ai" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-06-16", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Second TiRex version on GIFT-Eval (TiRex-2-Pretrained).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/spaces/Salesforce/GIFT-Eval", "https://huggingface.co/NX-AI/TiRex-2" ], "unverified": [] }, { "id": "nyu-langone-nyutron", "name": "NYUTron", "level": "model", "family_id": null, "developer": [ "nyu-langone" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": "BERT-style clinical encoder trained on 4.1B words of clinical notes", "params_total": 109000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-06-07", "release_date_basis": "notes", "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [ "internal only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "NYU Langone clinical language model for EHR prediction. Readmission AUC 79.87% (temporal test) and 78.70% (prospective); in-hospital mortality AUC 94.9%. Deployed through the EHR over HL7 FHIR from January to April 2022.", "verified_on": "2026-10-10", "sources": [ "https://www.nature.com/articles/s41586-023-06160-y" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "odyssey-2-max", "name": "Odyssey-2 Max", "level": "model", "family_id": null, "developer": [ "odyssey" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-06-17", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Odyssey general world model. Access method not stated.", "verified_on": "2026-10-10", "sources": [ "https://noqta.tn/en/news/odyssey-world-models-310m-series-b-2026" ], "unverified": [] }, { "id": "odyssey-agora-1", "name": "Agora-1", "level": "model", "family_id": null, "developer": [ "odyssey" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-06-17", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Odyssey multi-agent world model. Access method not stated.", "verified_on": "2026-10-10", "sources": [ "https://noqta.tn/en/news/odyssey-world-models-310m-series-b-2026" ], "unverified": [] }, { "id": "odyssey-starchild-1", "name": "Starchild-1", "level": "model", "family_id": null, "developer": [ "odyssey" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-06-17", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Odyssey model described by the vendor as the first real-time multimodal world model. Access method not stated.", "verified_on": "2026-10-10", "sources": [ "https://noqta.tn/en/news/odyssey-world-models-310m-series-b-2026" ], "unverified": [] }, { "id": "ola-krutrim-krutrim", "name": "Krutrim", "level": "model", "family_id": null, "developer": [ "ola-krutrim" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Not on AA.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/leaderboards/models" ], "unverified": [ "status" ] }, { "id": "omnianomaly", "name": "OmniAnomaly", "level": "model", "family_id": null, "developer": [], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Multivariate anomaly detector listed on TSB-AD (not a GNN).", "verified_on": "2026-10-10", "sources": [ "https://github.com/TheDatumOrg/TSB-AD" ], "unverified": [] }, { "id": "openai-chatgpt-4o-latest-2025-03", "name": "GPT-4o (March 2025, chatgpt-4o-latest)", "level": "variant", "family_id": "openai-gpt-4o-2025-03", "developer": [ "openai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 128000, "release_date": "2025-03-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "Max output 16,384 tokens; knowledge cutoff 1 October 2023", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "API alias chatgpt-4o-latest pointing to the GPT-4o snapshot used in ChatGPT; OpenAI states it has been deprecated and removed from the API.", "verified_on": "2026-10-11", "sources": [ "https://developers.openai.com/api/docs/models/chatgpt-4o-latest", "https://artificialanalysis.ai/models/gpt-4o-chatgpt-03-25", "https://community.openai.com/t/chatgpt-release-notes-2025-march-27-gpt-4o-a-new-update/1153887" ], "unverified": [] }, { "id": "openai-gpt-3", "name": "GPT-3 175B", "level": "model", "family_id": null, "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": 175000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Model used as the full fine-tuning baseline in the LoRA paper.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2106.09685" ], "unverified": [] }, { "id": "openai-gpt-3-5-turbo-0613", "name": "GPT-3.5 Turbo (0613)", "level": "model", "family_id": null, "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "Deprecation announced 2023-11-06; shut down 2024-09-13; replacement gpt-3.5-turbo", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "shut down", "summary": "gpt-3.5-turbo-0613 was a dated snapshot of the OpenAI GPT-3.5 Turbo chat model, shut down on 13 September 2024.", "verified_on": "2026-10-11", "sources": [ "https://developers.openai.com/api/docs/deprecations", "https://developers.openai.com/api/docs/models/gpt-3.5-turbo", "https://artificialanalysis.ai/models/gpt-35-turbo" ], "unverified": [] }, { "id": "openai-gpt-4", "name": "GPT-4", "level": "model", "family_id": null, "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "OpenAI model used as a comparison point in GAIA and in a finance-task study against BloombergGPT.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2311.12983", "https://arxiv.org/html/2305.05862v2" ], "unverified": [] }, { "id": "openai-gpt-4-1", "name": "GPT-4.1", "level": "model", "family_id": "openai-gpt-4-1-family", "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1047576, "release_date": "2025-04-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "32,768 max output tokens; snapshot gpt-4.1-2025-04-14", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenAI non-reasoning model for instruction following and tool calling with a context window of about one million tokens.", "verified_on": "2026-10-11", "sources": [ "https://developers.openai.com/api/docs/models/gpt-4.1.md", "https://developers.openai.com/api/docs/changelog.md" ], "unverified": [] }, { "id": "openai-gpt-4-1-family", "name": "GPT-4.1", "level": "family", "family_id": null, "developer": [ "openai" ], "model_class": "multimodal LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-04-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "OpenAI GPT-4.1 model line of gpt-4.1, gpt-4.1-mini and gpt-4.1-nano, added to the API on 14 April 2025.", "verified_on": "2026-10-11", "sources": [ "https://developers.openai.com/api/docs/changelog.md", "https://developers.openai.com/api/docs/models/gpt-4.1.md" ], "unverified": [] }, { "id": "openai-gpt-4-1-mini", "name": "GPT-4.1 mini", "level": "model", "family_id": "openai-gpt-4-1-family", "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1047576, "release_date": "2025-04-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "32,768 max output tokens; snapshot gpt-4.1-mini-2025-04-14", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Smaller, faster version of GPT-4.1 for instruction following and tool calling with a context window of about one million tokens.", "verified_on": "2026-10-11", "sources": [ "https://developers.openai.com/api/docs/models/gpt-4.1-mini.md", "https://developers.openai.com/api/docs/changelog.md" ], "unverified": [] }, { "id": "openai-gpt-4-1-nano", "name": "GPT-4.1 nano", "level": "model", "family_id": "openai-gpt-4-1-family", "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1047576, "release_date": "2025-04-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "32,768 max output tokens; snapshot gpt-4.1-nano-2025-04-14", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "Fastest and most cost-efficient version of GPT-4.1 with a context window of about one million tokens. OpenAI deprecated it with API shutdown set for 23 October 2026.", "verified_on": "2026-10-11", "sources": [ "https://developers.openai.com/api/docs/models/gpt-4.1-nano.md", "https://developers.openai.com/api/docs/changelog.md", "https://developers.openai.com/api/docs/deprecations.md" ], "unverified": [] }, { "id": "openai-gpt-4o-2025-03", "name": "GPT-4o (March 2025 update)", "level": "model", "family_id": null, "developer": [ "openai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 128000, "release_date": "2025-03-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "March 2025 update of GPT-4o in ChatGPT that follows instructions more accurately and handles coding more smoothly.", "verified_on": "2026-10-11", "sources": [ "https://community.openai.com/t/chatgpt-release-notes-2025-march-27-gpt-4o-a-new-update/1153887", "https://developers.openai.com/api/docs/models/chatgpt-4o-latest" ], "unverified": [] }, { "id": "openai-gpt-4o-mini", "name": "GPT-4o mini", "level": "model", "family_id": null, "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 128000, "release_date": "2024-07-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "16,384 max output tokens; snapshot gpt-4o-mini-2024-07-18", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fast, affordable small OpenAI model for focused tasks that accepts text and image input and returns text. OpenAI describes it as suited to fine-tuning and to distillation from larger models.", "verified_on": "2026-10-11", "sources": [ "https://developers.openai.com/api/docs/models/gpt-4o-mini.md", "https://developers.openai.com/api/docs/changelog.md" ], "unverified": [] }, { "id": "openai-gpt-4o-mini-tts", "name": "gpt-4o-mini-tts", "level": "model", "family_id": null, "developer": [ "openai" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Indonesian listed among languages that perform well, though voices are optimized for English", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenAI TTS model.", "verified_on": "2026-10-10", "sources": [ "https://developers.openai.com/api/docs/models", "https://developers.openai.com/api/docs/guides/text-to-speech" ], "unverified": [ "indonesian_measured" ] }, { "id": "openai-gpt-5", "name": "GPT-5", "level": "family", "family_id": null, "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-08-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "consumer app" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenAI's GPT-5.x line (GPT-5.2, 5.4, 5.5, 5.6 and Codex variants), predecessor of GPT-6.", "verified_on": "2026-10-10", "sources": [ "https://developers.openai.com/api/docs/models", "https://www.heyuan110.com/posts/ai/2026-07-10-gpt-5-6-general-availability/", "https://openai.com/index/introducing-gpt-5/", "https://cognite.com/en/atlas-ai-slm-llm-benchmark-report", "https://developers.openai.com/api/docs/models/gpt-5" ], "unverified": [ "indonesian_measured" ] }, { "id": "openai-gpt-5-1", "name": "GPT-5.1", "level": "model", "family_id": "openai-gpt-5", "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 400000, "release_date": "2025-11-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "128,000 max output tokens; knowledge cutoff Sep 30, 2024; reasoning effort none, low, medium or high", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "OpenAI flagship GPT-5 family model for coding and agentic tasks with configurable reasoning effort. OpenAI deprecated it on 1 October 2026 with API removal set for 1 April 2027.", "verified_on": "2026-10-11", "sources": [ "https://developers.openai.com/api/docs/models/gpt-5.1.md", "https://developers.openai.com/api/docs/changelog.md", "https://developers.openai.com/api/docs/deprecations.md" ], "unverified": [] }, { "id": "openai-gpt-5-1-codex", "name": "GPT-5.1-Codex", "level": "model", "family_id": "openai-gpt-5", "developer": [ "openai" ], "model_class": "code", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 400000, "release_date": "2025-11-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "128,000 max output tokens; Responses API only", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "shut down", "summary": "A version of GPT-5.1 optimized for agentic coding in Codex or similar environments. OpenAI shut it down in the API on 23 July 2026.", "verified_on": "2026-10-11", "sources": [ "https://developers.openai.com/api/docs/models/gpt-5.1-codex.md", "https://developers.openai.com/api/docs/changelog.md", "https://developers.openai.com/api/docs/deprecations.md" ], "unverified": [] }, { "id": "openai-gpt-5-1-codex-max", "name": "GPT-5.1-Codex-Max", "level": "model", "family_id": "openai-gpt-5", "developer": [ "openai" ], "model_class": "code", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 400000, "release_date": "2025-12-04", "release_date_basis": "inferred from page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "128,000 max output tokens; Responses API only. Release date is the API changelog date; the openai.com announcement page refused automated access", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "shut down", "summary": "A version of GPT-5.1-Codex optimized for long-running, long-horizon agentic coding tasks. OpenAI shut it down in the API on 23 July 2026.", "verified_on": "2026-10-11", "sources": [ "https://developers.openai.com/api/docs/models/gpt-5.1-codex-max.md", "https://developers.openai.com/api/docs/changelog.md", "https://developers.openai.com/api/docs/deprecations.md" ], "unverified": [] }, { "id": "openai-gpt-5-1-codex-mini", "name": "GPT-5.1-Codex-Mini", "level": "model", "family_id": "openai-gpt-5", "developer": [ "openai" ], "model_class": "code", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 400000, "release_date": "2025-11-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "128,000 max output tokens; knowledge cutoff Sep 30, 2024", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "shut down", "summary": "A smaller, more cost-effective and less capable version of GPT-5.1-Codex for coding tasks. OpenAI shut it down in the API on 23 July 2026.", "verified_on": "2026-10-11", "sources": [ "https://developers.openai.com/api/docs/models/gpt-5.1-codex-mini.md", "https://developers.openai.com/api/docs/changelog.md", "https://developers.openai.com/api/docs/deprecations.md" ], "unverified": [] }, { "id": "openai-gpt-5-2", "name": "GPT-5.2", "level": "model", "family_id": "openai-gpt-5", "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenAI model that OpenAI for Healthcare (8 Jan 2026) runs on; no numeric HealthBench score is published for it.", "verified_on": "2026-10-10", "sources": [ "https://developers.openai.com/api/docs/models", "https://openai.com/index/openai-for-healthcare/" ], "unverified": [ "indonesian_measured" ] }, { "id": "openai-gpt-5-3-codex", "name": "GPT-5.3-Codex", "level": "model", "family_id": "openai-gpt-5", "developer": [ "openai" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-02-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Codex-tuned GPT-5.x model for the Codex app, CLI, IDE and web; still GA in Copilot. By Oct 2026 Codex runs on GPT-5.6/6 general models.", "verified_on": "2026-10-10", "sources": [ "https://openai.com/index/introducing-gpt-5-3-codex", "https://docs.github.com/en/copilot/reference/ai-models/supported-models", "https://developers.openai.com/api/docs/models", "https://www.heyuan110.com/posts/ai/2026-07-10-gpt-5-6-general-availability/" ], "unverified": [] }, { "id": "openai-gpt-5-4", "name": "GPT-5.4", "level": "model", "family_id": "openai-gpt-5", "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenAI GPT-5.x model family, GA in GitHub Copilot.", "verified_on": "2026-10-10", "sources": [ "https://developers.openai.com/api/docs/models", "https://docs.github.com/en/copilot/reference/ai-models/supported-models", "https://www.datalearner.com/en/leaderboards/external/aa-quality-index" ], "unverified": [] }, { "id": "openai-gpt-5-5", "name": "GPT-5.5", "level": "model", "family_id": "openai-gpt-5", "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": "2026-04-23", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenAI GPT-5.x model, GA in GitHub Copilot.", "verified_on": "2026-10-10", "sources": [ "https://rits.shanghai.nyu.edu/ai/openai-releases-gpt-5-5-agentic-coding-ceiling-tops-14-benchmarks/", "https://docs.github.com/en/copilot/reference/ai-models/supported-models", "https://www.benchlm.ai/benchmarks/terminal-bench-2", "https://benchlm.ai/best/long-context", "https://www.abridge.com/", "https://capitalandcompute.net/ai-benchmarks/financebench/" ], "unverified": [ "indonesian_measured" ] }, { "id": "openai-gpt-5-6-cyber", "name": "GPT-5.6 Cyber", "level": "model", "family_id": "openai-gpt-5", "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Specialised GPT-5.6 sibling listed in OpenAI docs.", "verified_on": "2026-10-10", "sources": [ "https://developers.openai.com/api/docs/models" ], "unverified": [] }, { "id": "openai-gpt-5-6-luna", "name": "GPT-5.6 Luna", "level": "model", "family_id": "openai-gpt-5", "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": "2026-07-09", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Low-cost tier of the GPT-5.6 trio.", "verified_on": "2026-10-10", "sources": [ "https://www.heyuan110.com/posts/ai/2026-07-10-gpt-5-6-general-availability/" ], "unverified": [] }, { "id": "openai-gpt-5-6-sol", "name": "GPT-5.6 Sol", "level": "model", "family_id": "openai-gpt-5", "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": "2026-07-09", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "GA 2026-07-09; partner launch 2026-06-26. Max output 128K.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Top tier of the GPT-5.6 trio (July 2026); available in Codex, Copilot and Kiro.", "verified_on": "2026-10-10", "sources": [ "https://www.heyuan110.com/posts/ai/2026-07-10-gpt-5-6-general-availability/", "https://docs.github.com/en/copilot/reference/ai-models/supported-models", "https://kiro.dev/docs/models/", "https://www.benchlm.ai/benchmarks/frontiermath" ], "unverified": [] }, { "id": "openai-gpt-5-6-terra", "name": "GPT-5.6 Terra", "level": "model", "family_id": "openai-gpt-5", "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": "2026-07-09", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Middle tier of the GPT-5.6 trio.", "verified_on": "2026-10-10", "sources": [ "https://www.heyuan110.com/posts/ai/2026-07-10-gpt-5-6-general-availability/", "https://developers.openai.com/api/docs/models" ], "unverified": [] }, { "id": "openai-gpt-5-mini", "name": "GPT-5 mini", "level": "model", "family_id": "openai-gpt-5", "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small GPT-5 model, GA in GitHub Copilot.", "verified_on": "2026-10-10", "sources": [ "https://docs.github.com/en/copilot/reference/ai-models/supported-models", "https://cognite.com/en/atlas-ai-slm-llm-benchmark-report" ], "unverified": [ "indonesian_measured" ] }, { "id": "openai-gpt-6", "name": "GPT-6", "level": "family", "family_id": null, "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenAI's GPT-6 family; flagship GPT-6 Astra for complex reasoning and coding, with Sol and Luna tiers.", "verified_on": "2026-10-10", "sources": [ "https://developers.openai.com/api/docs/models", "https://developers.openai.com/api/docs/models" ], "unverified": [] }, { "id": "openai-gpt-6-1-astra", "name": "GPT-6.1 Astra", "level": "model", "family_id": "openai-gpt-6", "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-09-28", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Reportedly dropped 2026-09-28 (secondary report, unverified).", "verified_on": "2026-10-10", "sources": [ "https://slash-digital.io/en/insights/ai-models-2026/" ], "unverified": [ "release_date", "status" ] }, { "id": "openai-gpt-6-1-sol", "name": "GPT-6.1 Sol", "level": "model", "family_id": "openai-gpt-6", "developer": [ "openai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1050000, "release_date": "2026-09-29", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "Max output 128K. Vendor cloud/API only.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "GPT-6.1 mid tier model with effort levels low to max and tool support (functions, web search, file search, computer use).", "verified_on": "2026-10-10", "sources": [ "https://developers.openai.com/api/docs/models", "https://llm-stats.com/llm-updates", "https://slash-digital.io/en/insights/ai-models-2026/", "https://developers.openai.com/api/docs/models" ], "unverified": [] }, { "id": "openai-gpt-6-astra", "name": "GPT-6 Astra", "level": "model", "family_id": "openai-gpt-6", "developer": [ "openai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed; pre-trained on more than 100,000 GPUs at Stargate Texas (secondary, citing Axios)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1050000, "release_date": "2026-09-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "consumer app", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": "Vendor cloud/API only (T1). Max output 128K. Staged rollout from 3 Sep 2026; second OpenAI post 9 Sep 2026.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Flagship of the GPT-6 family for complex reasoning and coding; text and image in, text out, with function calling, web search, file search and computer use. Available via OpenAI API, ChatGPT, GitHub Copilot.", "verified_on": "2026-10-10", "sources": [ "https://developers.openai.com/api/docs/models", "https://openai.com/index/gpt-6-astra/", "https://openai.com/news/rss.xml", "https://pasqualepillitteri.it/en/news/14246/openai-launches-gpt-6-astra-pricing-benchmarks", "https://docs.apiyi.com/en/news/gpt-6-astra-launch", "https://infoq.com/news/2026/09/openai-gpt6-astra", "https://docs.github.com/en/copilot/reference/ai-models/supported-models", "https://www.benchlm.ai/benchmarks/gpqa", "https://epoch.ai/benchmarks" ], "unverified": [] }, { "id": "openai-gpt-6-luna", "name": "GPT-6 Luna", "level": "model", "family_id": "openai-gpt-6", "developer": [ "openai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1050000, "release_date": "2026-09-22", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "Max output 128K.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Lowest-cost GPT-6 tier with effort levels none to max (reported 22 or 23 Sep 2026).", "verified_on": "2026-10-10", "sources": [ "https://www.deccanchronicle.com/technology/openai-expands-gpt-6-lineup-with-cheaper-sol-and-luna-models-1989855", "https://developers.openai.com/api/docs/models", "https://developers.openai.com/api/docs/models" ], "unverified": [] }, { "id": "openai-gpt-6-sol", "name": "GPT-6 Sol", "level": "model", "family_id": "openai-gpt-6", "developer": [ "openai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1050000, "release_date": "2026-09-22", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Cheaper GPT-6 tier announced alongside GPT-6 Luna (reported 22 or 23 Sep 2026).", "verified_on": "2026-10-10", "sources": [ "https://www.deccanchronicle.com/technology/openai-expands-gpt-6-lineup-with-cheaper-sol-and-luna-models-1989855", "https://developers.openai.com/api/docs/models" ], "unverified": [ "status" ] }, { "id": "openai-gpt-image", "name": "GPT Image", "level": "family", "family_id": null, "developer": [ "openai" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenAI's image generation model line (gpt-image-1.5, gpt-image-2, gpt-image-2.5).", "verified_on": "2026-10-10", "sources": [ "https://docs.apiyi.com/en/news/gpt-image-2-5-launch", "https://developers.openai.com/api/docs/models" ], "unverified": [] }, { "id": "openai-gpt-image-1-5", "name": "GPT Image 1.5", "level": "model", "family_id": "openai-gpt-image", "developer": [ "openai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-12", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenAI image model (gpt-image-1.5) released Dec 2025.", "verified_on": "2026-10-10", "sources": [ "https://docs.apiyi.com/en/news/gpt-image-2-5-launch", "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text" ], "unverified": [] }, { "id": "openai-gpt-image-2", "name": "GPT Image 2", "level": "model", "family_id": "openai-gpt-image", "developer": [ "openai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenAI image model (gpt-image-2) released Apr 2026, predecessor of GPT Image 2.5.", "verified_on": "2026-10-10", "sources": [ "https://docs.apiyi.com/en/news/gpt-image-2-5-launch", "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text", "https://arena.ai/leaderboard/text-to-image" ], "unverified": [] }, { "id": "openai-gpt-image-2-5-flare", "name": "GPT Image 2.5 Flare", "level": "model", "family_id": "openai-gpt-image", "developer": [ "openai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-09-08", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "consumer app" ], "deployment_tiers": [], "footprint_note": "Cloud only; API ID gpt-image-2.5-flare with dated snapshot 2026-09-08", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenAI image generation model variant with up to 50% lower latency than Sunburst, available in the API and ChatGPT.", "verified_on": "2026-10-10", "sources": [ "https://docs.apiyi.com/en/news/gpt-image-2-5-launch", "https://developers.openai.com/api/docs/models", "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text", "https://arena.ai/leaderboard/text-to-image" ], "unverified": [] }, { "id": "openai-gpt-image-2-5-sunburst", "name": "GPT Image 2.5 Sunburst", "level": "model", "family_id": "openai-gpt-image", "developer": [ "openai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-09-08", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "consumer app" ], "deployment_tiers": [], "footprint_note": "Cloud only; API ID gpt-image-2.5-sunburst with dated snapshot 2026-09-08", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenAI image generation and editing model tuned for quality, available in the API and ChatGPT. Priced like gpt-image-2 at US$30 per 1M output image tokens.", "verified_on": "2026-10-10", "sources": [ "https://docs.apiyi.com/en/news/gpt-image-2-5-launch", "https://developers.openai.com/api/docs/models", "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text", "https://arena.ai/leaderboard/text-to-image" ], "unverified": [] }, { "id": "openai-gpt-live-1", "name": "gpt-live-1", "level": "model", "family_id": null, "developer": [ "openai" ], "model_class": "speech synthesis", "modalities_in": [ "text", "audio" ], "modalities_out": [ "audio", "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Realtime voice model in OpenAI's API.", "verified_on": "2026-10-10", "sources": [ "https://developers.openai.com/api/docs/models" ], "unverified": [] }, { "id": "openai-gpt-oss", "name": "gpt-oss", "level": "family", "family_id": null, "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-08-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenAI open-weight line; not checked in the research notes (unverified).", "verified_on": "2026-10-10", "sources": [ "https://developers.openai.com/api/docs/models", "https://huggingface.co/openai/gpt-oss-120b", "https://huggingface.co/openai", "https://huggingface.co/openai/gpt-oss-safeguard-20b", "https://openai.com/index/introducing-gpt-oss/" ], "unverified": [ "status" ] }, { "id": "openai-gpt-oss-120b", "name": "gpt-oss-120b", "level": "model", "family_id": "openai-gpt-oss", "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "MoE, MXFP4", "params_total": 117000000000, "params_active": 5100000000, "params_checkpoint": null, "params_note": null, "context_length": 131000, "release_date": "2025-08-05", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": "Single 80 GB GPU", "runtimes": [], "languages_note": null, "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "in-region Jakarta", "status": "active", "summary": "OpenAI's 117B open-weight MoE text model (Apache 2.0 plus harmony format requirement). Adapters 232, finetunes 122, merges 1, quants 134; 3.97M downloads/month.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/openai/gpt-oss-120b", "https://codersera.com/blog/open-source-llms-landscape-2026/", "https://leaderboard.sea-lion.ai/detailed/SEA", "https://leaderboard.sea-lion.ai/detailed/ID", "https://openai.com/index/introducing-gpt-oss/", "https://docs.aws.amazon.com/bedrock/latest/userguide/models-region-compatibility.md" ], "unverified": [], "hosting_provider": "Amazon Bedrock (ap-southeast-3 Jakarta)", "hosting_url": "https://docs.aws.amazon.com/bedrock/latest/userguide/models-region-compatibility.md", "indonesian_measured": [ { "benchmark": "SEA-HELM Indonesian leaderboard (ID aggregate, first score column of the Indonesian Competency and Task Scores table)", "score": "72.28 (95% CI -1.35 to +1.33)", "metric": "score 0 to 100, average of 2000 bootstraps with 95% CI", "run_by": "AI Singapore (SEA-HELM leaderboard maintainer)", "url": "https://leaderboard.sea-lion.ai/detailed/ID" } ] }, { "id": "openai-gpt-oss-20b", "name": "gpt-oss-20b", "level": "model", "family_id": "openai-gpt-oss", "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 21000000000, "params_active": 3600000000, "params_checkpoint": null, "params_note": null, "context_length": 131000, "release_date": "2025-08-05", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": "16 GB consumer GPU", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenAI's 21B open-weight MoE text model; base of gpt-oss-safeguard-20b and GPT-OSS-Swallow-20B.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/openai/gpt-oss-120b", "https://huggingface.co/openai", "https://huggingface.co/openai/gpt-oss-safeguard-20b", "https://mlcommons.org/benchmarks/training/", "https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct", "https://openai.com/index/introducing-gpt-oss/", "https://gigazine.net/gsc_news/en/20250806-openai-gpt-oss", "https://huggingface.co/api/models/openai/gpt-oss-20b", "https://github.com/openai/gpt-oss" ], "unverified": [] }, { "id": "openai-gpt-oss-safeguard-120b", "name": "gpt-oss-safeguard-120b", "level": "model", "family_id": "openai-gpt-oss", "developer": [ "openai" ], "model_class": "guard", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 117000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Not stated", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenAI policy-as-prompt reasoning classifier with configurable reasoning effort.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/openai", "https://huggingface.co/openai/gpt-oss-safeguard-20b" ], "unverified": [] }, { "id": "openai-gpt-oss-safeguard-20b", "name": "gpt-oss-safeguard-20b", "level": "model", "family_id": "openai-gpt-oss", "developer": [ "openai" ], "model_class": "guard", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 21000000000, "params_active": 3600000000, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Not stated", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenAI policy-as-prompt reasoning classifier with configurable reasoning effort. 20b refreshed Jan 2026; finetunes 7, merges 2, quants 14.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/openai", "https://huggingface.co/openai/gpt-oss-safeguard-20b" ], "unverified": [] }, { "id": "openai-gpt-realtime-1-5", "name": "gpt-realtime-1.5", "level": "model", "family_id": null, "developer": [ "openai" ], "model_class": "speech synthesis", "modalities_in": [ "text", "audio" ], "modalities_out": [ "audio", "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Realtime voice model in OpenAI's API.", "verified_on": "2026-10-10", "sources": [ "https://developers.openai.com/api/docs/models" ], "unverified": [] }, { "id": "openai-gpt-realtime-2", "name": "gpt-realtime-2", "level": "model", "family_id": null, "developer": [ "openai" ], "model_class": "speech synthesis", "modalities_in": [ "text", "audio" ], "modalities_out": [ "audio", "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Realtime voice model in OpenAI's API.", "verified_on": "2026-10-10", "sources": [ "https://developers.openai.com/api/docs/models" ], "unverified": [] }, { "id": "openai-gpt-realtime-2-1", "name": "gpt-realtime-2.1", "level": "model", "family_id": null, "developer": [ "openai" ], "model_class": "speech synthesis", "modalities_in": [ "text", "audio" ], "modalities_out": [ "audio", "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Realtime voice model in OpenAI's API (also 2.1-mini).", "verified_on": "2026-10-10", "sources": [ "https://developers.openai.com/api/docs/models" ], "unverified": [] }, { "id": "openai-gpt-realtime-2-1-mini", "name": "gpt-realtime-2.1-mini", "level": "model", "family_id": null, "developer": [ "openai" ], "model_class": "speech synthesis", "modalities_in": [ "text", "audio" ], "modalities_out": [ "audio", "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Realtime voice model in OpenAI's API.", "verified_on": "2026-10-10", "sources": [ "https://developers.openai.com/api/docs/models" ], "unverified": [] }, { "id": "openai-gpt-realtime-mini", "name": "gpt-realtime-mini", "level": "model", "family_id": null, "developer": [ "openai" ], "model_class": "speech synthesis", "modalities_in": [ "text", "audio" ], "modalities_out": [ "audio", "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "Deprecated OpenAI realtime voice model.", "verified_on": "2026-10-10", "sources": [ "https://developers.openai.com/api/docs/models" ], "unverified": [] }, { "id": "openai-gpt-realtime-translate", "name": "gpt-realtime-translate", "level": "model", "family_id": null, "developer": [ "openai" ], "model_class": "speech synthesis", "modalities_in": [ "text", "audio" ], "modalities_out": [ "audio", "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Realtime translation voice model in OpenAI's API.", "verified_on": "2026-10-10", "sources": [ "https://developers.openai.com/api/docs/models" ], "unverified": [] }, { "id": "openai-o-series", "name": "o-series", "level": "family", "family_id": null, "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "OpenAI reasoning series described as legacy.", "verified_on": "2026-10-10", "sources": [ "https://developers.openai.com/api/docs/models" ], "unverified": [] }, { "id": "openai-o1", "name": "o1", "level": "model", "family_id": null, "developer": [ "openai" ], "model_class": "reasoning", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-09-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "OpenAI reasoning model whose performance OpenAI states improves with both train-time and test-time compute.", "verified_on": "2026-10-10", "sources": [ "https://openai.com/index/learning-to-reason-with-llms/" ], "unverified": [] }, { "id": "openai-o1-pro", "name": "o1-pro", "level": "model", "family_id": "openai-o-series", "developer": [ "openai" ], "model_class": "reasoning", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 200000, "release_date": "2025-03-19", "release_date_basis": "inferred from page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "Responses API and Batch only; max output 100,000 tokens; knowledge cutoff 1 October 2023", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Version of OpenAI o1 that uses more compute to think harder, offered in the API as snapshot o1-pro-2025-03-19.", "verified_on": "2026-10-11", "sources": [ "https://developers.openai.com/api/docs/models/o1-pro", "https://openrouter.ai/api/v1/models" ], "unverified": [] }, { "id": "openai-o3", "name": "o3", "level": "model", "family_id": "openai-o-series", "developer": [ "openai" ], "model_class": "reasoning", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 200000, "release_date": "2025-04-16", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "100,000 max output tokens; knowledge cutoff Jun 01, 2024; snapshot o3-2025-04-16", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "OpenAI o-series reasoning model for math, science, coding and visual reasoning, succeeded by GPT-5. The o3-2025-04-16 snapshot is deprecated with API removal set for 11 December 2026.", "verified_on": "2026-10-11", "sources": [ "https://developers.openai.com/api/docs/models/o3.md", "https://developers.openai.com/api/docs/changelog.md", "https://developers.openai.com/api/docs/deprecations.md" ], "unverified": [] }, { "id": "openai-o3-mini", "name": "o3-mini", "level": "model", "family_id": "openai-o-series", "developer": [ "openai" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 200000, "release_date": "2025-01-31", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "100,000 max output tokens; knowledge cutoff Oct 01, 2023", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "Small OpenAI reasoning model optimized for science, math and coding at the cost and latency of o1-mini. OpenAI deprecated it with API shutdown set for 23 October 2026.", "verified_on": "2026-10-11", "sources": [ "https://developers.openai.com/api/docs/models/o3-mini.md", "https://developers.openai.com/api/docs/changelog.md", "https://developers.openai.com/api/docs/deprecations.md" ], "unverified": [] }, { "id": "openai-o3-pro", "name": "o3-pro", "level": "model", "family_id": "openai-o-series", "developer": [ "openai" ], "model_class": "reasoning", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 200000, "release_date": "2025-06-10", "release_date_basis": "inferred from page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "Responses API and Batch only; max output 100,000 tokens; knowledge cutoff 1 June 2024", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Version of OpenAI o3 that uses more compute for better responses, available through the Responses API (snapshot o3-pro-2025-06-10).", "verified_on": "2026-10-11", "sources": [ "https://developers.openai.com/api/docs/models/o3-pro", "https://openrouter.ai/api/v1/models" ], "unverified": [] }, { "id": "openai-o4-mini", "name": "o4-mini", "level": "model", "family_id": "openai-o-series", "developer": [ "openai" ], "model_class": "reasoning", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 200000, "release_date": "2025-04-16", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "100,000 max output tokens; snapshot o4-mini-2025-04-16", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "Small, fast o-series reasoning model optimized for coding and visual tasks, succeeded by GPT-5 mini. OpenAI deprecated it with API shutdown set for 23 October 2026.", "verified_on": "2026-10-11", "sources": [ "https://developers.openai.com/api/docs/models/o4-mini.md", "https://developers.openai.com/api/docs/changelog.md", "https://developers.openai.com/api/docs/deprecations.md" ], "unverified": [] }, { "id": "openai-omni-moderation-latest", "name": "OpenAI Moderation (omni-moderation-latest)", "level": "model", "family_id": null, "developer": [ "openai" ], "model_class": "guard", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Not stated on the page", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenAI's free moderation API for text and image (no audio) across 13 categories; the model is upgraded continuously.", "verified_on": "2026-10-10", "sources": [ "https://developers.openai.com/api/docs/guides/moderation" ], "unverified": [] }, { "id": "openai-privacy-filter", "name": "OpenAI privacy-filter", "level": "model", "family_id": null, "developer": [ "openai" ], "model_class": "guard", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 1000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04-22", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenAI 1B token-classification model for PII.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/openai", "https://huggingface.co/openai/gpt-oss-safeguard-20b" ], "unverified": [ "summary", "licence" ] }, { "id": "openai-sora", "name": "Sora", "level": "model", "family_id": "openai-sora-family", "developer": [ "openai" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Original Sora model, listed on VBench-2.0 at 480p.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/spaces/Vchitect/VBench_Leaderboard" ], "unverified": [] }, { "id": "openai-sora-2", "name": "Sora 2", "level": "model", "family_id": "openai-sora-family", "developer": [ "openai" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "shut down", "summary": "OpenAI video model; app closed 26 Apr 2026 and API ended 24 Sep 2026 after a shutdown announced 24 Mar 2026.", "verified_on": "2026-10-10", "sources": [ "https://zilliz.com/ai-faq/what-is-the-sora-shutdown-timeline", "https://developers.openai.com/api/docs/models" ], "unverified": [] }, { "id": "openai-sora-2-pro", "name": "Sora 2 Pro", "level": "variant", "family_id": "openai-sora-2", "developer": [ "openai" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "shut down", "summary": "Pro variant of Sora 2, still listed on LMArena after shutdown.", "verified_on": "2026-10-10", "sources": [ "https://zilliz.com/ai-faq/what-is-the-sora-shutdown-timeline", "https://arena.ai/leaderboard/text-to-video" ], "unverified": [] }, { "id": "openai-sora-family", "name": "Sora", "level": "family", "family_id": null, "developer": [ "openai" ], "model_class": "video generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "shut down", "summary": "OpenAI video generation line, discontinued in 2026.", "verified_on": "2026-10-10", "sources": [ "https://zilliz.com/ai-faq/what-is-the-sora-shutdown-timeline" ], "unverified": [] }, { "id": "openai-whisper", "name": "Whisper", "level": "family", "family_id": null, "developer": [ "openai" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "autoregressive", "architecture_detail": "Transformer encoder-decoder", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "99+ languages", "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenAI multilingual ASR family; root of many fine-tunes and included in the FunASR model zoo.", "verified_on": "2026-10-10", "sources": [ "https://www.gladia.io/blog/best-open-source-speech-to-text-models", "https://github.com/modelscope/FunASR", "https://arxiv.org/html/2410.08828v2" ], "unverified": [ "indonesian_measured" ] }, { "id": "openai-whisper-large-v3", "name": "Whisper large-v3", "level": "variant", "family_id": "openai-whisper", "developer": [ "openai" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "autoregressive", "architecture_detail": null, "params_total": 1550000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": "99+ languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Whisper large-v3; led the Open ASR Leaderboard in 2023.", "verified_on": "2026-10-10", "sources": [ "https://www.gladia.io/blog/best-open-source-speech-to-text-models", "https://select.keenable.ai/r/speech-recognition-accuracy-leaders-2022-2026/JbURQjlVooLQ6J10NkbRVw" ], "unverified": [ "release_date" ] }, { "id": "openai-whisper-large-v3-turbo", "name": "Whisper large-v3-turbo", "level": "variant", "family_id": "openai-whisper", "developer": [ "openai" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "autoregressive", "architecture_detail": null, "params_total": 809000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU", "CPU server" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Faster Whisper large-v3 variant; CPU-feasible.", "verified_on": "2026-10-10", "sources": [ "https://www.gladia.io/blog/best-open-source-speech-to-text-models" ], "unverified": [ "release_date" ] }, { "id": "openai-whisper-small", "name": "whisper-small", "level": "variant", "family_id": "openai-whisper", "developer": [ "openai" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "autoregressive", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-09-26", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small Whisper model evaluated zero-shot on the Indonesian IDSV test set.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2410.08828v2", "https://huggingface.co/openai/whisper-small" ], "unverified": [ "indonesia_hosting" ], "indonesian_measured": [ { "benchmark": "IDSV Indonesian speech-variability test set, all groups (whisper-small, not fine-tuned)", "score": "30.87", "metric": "WER %, lower is better", "run_by": "paper authors (arXiv 2410.08828)", "url": "https://arxiv.org/html/2410.08828v2" } ] }, { "id": "openai-whisper-tiny", "name": "whisper-tiny", "level": "variant", "family_id": "openai-whisper", "developer": [ "openai" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Tiny Whisper model converted by Systran to CTranslate2 as faster-whisper-tiny.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/Systran/faster-whisper-tiny" ], "unverified": [] }, { "id": "openbmb-bitcpm4-0-5b", "name": "BitCPM4 0.5B", "level": "variant", "family_id": "openbmb-minicpm4", "developer": [ "openbmb" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "0.5B ternary", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Ternary-weight BitCPM4 0.5B model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/openbmb/MiniCPM4.1-8B-Eagle3/blob/main/README.md" ], "unverified": [] }, { "id": "openbmb-bitcpm4-1b", "name": "BitCPM4 1B", "level": "variant", "family_id": "openbmb-minicpm4", "developer": [ "openbmb" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "1B ternary", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Ternary-weight BitCPM4 1B model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/openbmb/MiniCPM4.1-8B-Eagle3/blob/main/README.md" ], "unverified": [] }, { "id": "openbmb-minicpm-v-4-6-1-3b", "name": "MiniCPM-V 4.6 1.3B", "level": "model", "family_id": null, "developer": [ "openbmb" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "1.3B dense", "context_length": 262000, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": "BF16 at release. Not stated as on-device by the source; Ollama lists it as phone-oriented", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenBMB 1.3B vision-language model; highest open model under 2B on the AA Intelligence Index at 11 May 2026.", "verified_on": "2026-10-10", "sources": [ "https://www.artificialanalysis.ai/articles/openbmb-launches-minicpm-v-4-6-1-3b-instruct", "https://ollama.com/library" ], "unverified": [] }, { "id": "openbmb-minicpm3-1b", "name": "MiniCPM3-1B", "level": "model", "family_id": null, "developer": [ "openbmb" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "1B", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier MiniCPM model from which BitCPM4-1B is derived.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/openbmb/MiniCPM4.1-8B-Eagle3/blob/main/README.md" ], "unverified": [] }, { "id": "openbmb-minicpm4", "name": "MiniCPM4 / 4.1", "level": "family", "family_id": null, "developer": [ "openbmb" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-06-06", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "GPTQ, AWQ, GGUF, MLX variants; over 5x generation acceleration claimed on typical end-side chips (chips not named)", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenBMB on-device model family (MiniCPM4 6 Jun 2025, MiniCPM4.1 5 Sep 2025).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/openbmb/MiniCPM4.1-8B-Eagle3/blob/main/README.md", "https://cnb.cool/ai-models/openbmb/MiniCPM4-0.5B" ], "unverified": [] }, { "id": "openbmb-minicpm4-0-5b", "name": "MiniCPM4 0.5B", "level": "model", "family_id": "openbmb-minicpm4", "developer": [ "openbmb" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "0.5B, trained on 1T tokens", "context_length": null, "release_date": "2025-06-06", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MiniCPM4 0.5B on-device model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/openbmb/MiniCPM4.1-8B-Eagle3/blob/main/README.md", "https://cnb.cool/ai-models/openbmb/MiniCPM4-0.5B" ], "unverified": [] }, { "id": "openbmb-minicpm4-1-8b", "name": "MiniCPM4.1 8B", "level": "model", "family_id": "openbmb-minicpm4", "developer": [ "openbmb" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "8B", "context_length": null, "release_date": "2025-09-05", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MiniCPM4.1 8B update released 5 Sep 2025.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/openbmb/MiniCPM4.1-8B-Eagle3/blob/main/README.md" ], "unverified": [] }, { "id": "openbmb-minicpm4-8b", "name": "MiniCPM4 8B", "level": "model", "family_id": "openbmb-minicpm4", "developer": [ "openbmb" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "8B, trained on 8T tokens", "context_length": null, "release_date": "2025-06-06", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MiniCPM4 8B model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/openbmb/MiniCPM4.1-8B-Eagle3/blob/main/README.md" ], "unverified": [] }, { "id": "openbmb-voxcpm2", "name": "OpenBMB VoxCPM2", "level": "model", "family_id": null, "developer": [ "openbmb" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open TTS model trending on Hugging Face (2025 to 2026).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/models?pipeline_tag=text-to-speech&sort=trending", "https://huggingface.co/openbmb/VoxCPM2" ], "unverified": [] }, { "id": "opendatalab-mineru2-5", "name": "MinerU2.5 (MinerU-2.5)", "level": "model", "family_id": null, "developer": [ "opendatalab" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "autoregressive", "architecture_detail": null, "params_total": 1160000000, "params_active": null, "params_checkpoint": null, "params_note": "1.2B in the research notes; HF MinerU2.5-2509-1.2B reports 1.16B", "context_length": null, "release_date": "2025-09-17", "release_date_basis": "registry date", "weights_date": null, "openness": "open source", "licence": "agpl-3-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Document-parsing VLM from OpenDataLab (Shanghai AI Lab link unverified).", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2606.03264", "https://huggingface.co/api/models/opendatalab/MinerU2.5-2509-1.2B" ], "unverified": [] }, { "id": "opendatalab-mineru2-5-pro", "name": "MinerU2.5-Pro", "level": "model", "family_id": null, "developer": [ "opendatalab" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "autoregressive", "architecture_detail": "Compact VLM, coarse-to-fine", "params_total": 1200000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Compact document-parsing VLM, successor of MinerU 2.5.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2606.03264" ], "unverified": [ "developer", "release_date", "licence" ] }, { "id": "openfold-consortium-openfold3", "name": "OpenFold3 (preview)", "level": "model", "family_id": null, "developer": [ "openfold-consortium" ], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "other", "architecture_detail": "AlphaFold3 replication", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10-28", "release_date_basis": "secondary report", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": "AF3 replication by the OpenFold consortium (BMS, Novo Nordisk, Bayer, Roche, UCB etc.); claimed on par with AF3 for protein-ligand co-folding (company claim).", "verified_on": "2026-10-10", "sources": [ "https://www.rdworldonline.com/openfold3-debuts-as-apache-licensed-protein-co-folding-model-with-apheris-bringing-it-into-pharma-it/", "https://scalingbiotech.substack.com/p/open-source-protein-models-openfold" ], "unverified": [] }, { "id": "opengpt-x-teuken-7b-v0-6", "name": "Teuken-7B v0.6", "level": "model", "family_id": null, "developer": [ "opengpt-x" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 7000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-08", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": "24 EU languages; no Indonesian", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenGPT-X's 7B model for EU languages; no 2026 release found.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/openGPT-X" ], "unverified": [ "licence" ] }, { "id": "openmoss-moss-transcribe-diarize", "name": "MOSS-Transcribe-Diarize", "level": "model", "family_id": null, "developer": [ "openmoss" ], "model_class": "speaker", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Offline ASR with anonymous speaker labels; licence per its own card.", "verified_on": "2026-10-10", "sources": [ "https://github.com/modelscope/FunASR" ], "unverified": [] }, { "id": "orbital-materials-orb-v3", "name": "ORB v3", "level": "model", "family_id": null, "developer": [ "orbital-materials" ], "model_class": "materials", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "GNN", "architecture_detail": "Equivariant GNN interatomic potential (MLIP)", "params_total": 25500000, "params_active": null, "params_checkpoint": null, "params_note": "Training data MPtrj+Alex+OMat24", "context_length": null, "release_date": "2025-04-05", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Machine-learned interatomic potential on Matbench Discovery (rank 10 by CPS, date added 2025-04-05).", "verified_on": "2026-10-10", "sources": [ "https://matbench-discovery.materialsproject.org/" ], "unverified": [ "developer" ] }, { "id": "paano", "name": "PaAno", "level": "model", "family_id": null, "developer": [ "sungkyunkwan-university" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Time-series anomaly detector ranked 1 on TSB-AD (VUS-PR 0.53 U, 0.43 M) in its ICLR 2026 paper.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2602.01359", "https://arxiv.org/html/2602.01359" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "paige-virchow", "name": "Virchow", "level": "model", "family_id": null, "developer": [ "paige" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": "ViT-H/14 trained with DINOv2 self-supervision on about 1.5M H&E slides from MSKCC", "params_total": 632000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-07-22", "release_date_basis": "notes", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Pathology whole-slide image foundation model. Published in Nature Medicine on 22 Jul 2024 with pan-cancer specimen-level AUC 0.950.", "verified_on": "2026-10-10", "sources": [ "https://www.nature.com/articles/s41591-024-03141-0" ], "unverified": [] }, { "id": "pca", "name": "PCA", "level": "model", "family_id": null, "developer": [], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Statistical anomaly detection baseline on TSB-AD and SKAB.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2602.01359", "https://github.com/waico/SKAB" ], "unverified": [] }, { "id": "perpetualbooster", "name": "PerpetualBooster", "level": "model", "family_id": null, "developer": [], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "gradient boosting", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Boosting model on TabArena.", "verified_on": "2026-10-10", "sources": [ "https://benchmarklist.com/benchmarks/tabarena_all/" ], "unverified": [] }, { "id": "perplexity-sonar", "name": "Sonar", "level": "family", "family_id": null, "developer": [ "perplexity" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Sonar is the Perplexity family of search-grounded API models.", "verified_on": "2026-10-11", "sources": [ "https://docs.perplexity.ai/getting-started/models/models/sonar-pro", "https://openrouter.ai/api/v1/models" ], "unverified": [] }, { "id": "perplexity-sonar-pro", "name": "Sonar Pro", "level": "model", "family_id": "perplexity-sonar", "developer": [ "perplexity" ], "model_class": "LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 200000, "release_date": "2025-03-07", "release_date_basis": "registry date", "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "Perplexity search-grounded API model for multi-step queries; Perplexity ends Sonar Chat Completions support on 27 September 2026 and maps Sonar Pro to the Agent API fast preset.", "verified_on": "2026-10-11", "sources": [ "https://docs.perplexity.ai/getting-started/models/models/sonar-pro", "https://openrouter.ai/api/v1/models", "https://docs.perplexity.ai/getting-started/models" ], "unverified": [] }, { "id": "perplexity-sonar-reasoning-pro", "name": "Sonar Reasoning Pro", "level": "model", "family_id": "perplexity-sonar", "developer": [ "perplexity" ], "model_class": "reasoning", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 128000, "release_date": "2025-03-07", "release_date_basis": "registry date", "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "Perplexity search-grounded reasoning API model, described by OpenRouter as powered by DeepSeek R1; Perplexity ends Sonar Chat Completions support on 27 September 2026.", "verified_on": "2026-10-11", "sources": [ "https://docs.perplexity.ai/getting-started/models/models/sonar-reasoning-pro", "https://openrouter.ai/api/v1/models", "https://docs.perplexity.ai/getting-started/models" ], "unverified": [] }, { "id": "phikon", "name": "Phikon", "level": "model", "family_id": null, "developer": [ "owkin" ], "model_class": "vision", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-09-21", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Pathology model used as a comparison in the Virchow study (AUC 0.932); developer not stated in these sections.", "verified_on": "2026-10-10", "sources": [ "https://www.nature.com/articles/s41591-024-03141-0", "https://huggingface.co/owkin/phikon" ], "unverified": [] }, { "id": "physical-intelligence-pi", "name": "π (pi) models", "level": "family", "family_id": null, "developer": [ "physical-intelligence" ], "model_class": "robot policy", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Physical Intelligence vision-language-action models.", "verified_on": "2026-10-10", "sources": [ "https://www.pi.website/" ], "unverified": [] }, { "id": "physical-intelligence-pi-star-0-6", "name": "π*0.6", "level": "model", "family_id": "physical-intelligence-pi", "developer": [ "physical-intelligence" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": null, "architecture_detail": "VLA improved through RL from experience (Recap)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-11-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "internal only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Physical Intelligence VLA improved with RL from experience; not open.", "verified_on": "2026-10-10", "sources": [ "https://www.pi.website/" ], "unverified": [] }, { "id": "physical-intelligence-pi0", "name": "π0", "level": "model", "family_id": "physical-intelligence-pi", "developer": [ "physical-intelligence" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": "flow matching", "architecture_detail": "Flow-matching VLA", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-10-31", "release_date_basis": "announcement page", "weights_date": "2025-02-04", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Physical Intelligence flow-matching VLA, open-sourced in the openpi repo (Apache-2.0 plus a separate Gemma licence file).", "verified_on": "2026-10-10", "sources": [ "https://www.pi.website/", "https://github.com/Physical-Intelligence/openpi" ], "unverified": [] }, { "id": "physical-intelligence-pi0-5", "name": "π0.5", "level": "model", "family_id": "physical-intelligence-pi", "developer": [ "physical-intelligence" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": null, "architecture_detail": "Upgraded π0 with knowledge insulation", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-04-22", "release_date_basis": "announcement page", "weights_date": "2025-09", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "π0 upgrade for open-world mobile manipulation in unseen homes; openpi calls its LIBERO checkpoint state-of-the-art without a number.", "verified_on": "2026-10-10", "sources": [ "https://www.pi.website/", "https://github.com/Physical-Intelligence/openpi" ], "unverified": [] }, { "id": "physical-intelligence-pi0-7", "name": "π0.7", "level": "model", "family_id": "physical-intelligence-pi", "developer": [ "physical-intelligence" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04-16", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "research only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "VLA with compositional generalisation from language; architecture, base model and openness not disclosed.", "verified_on": "2026-10-10", "sources": [ "https://ai2.work/blog/physical-intelligence-nears-11b-valuation-as-pi-0-7-learns-what-it-was-never", "https://www.tmtpost.com/agent/ai-article?id=14956" ], "unverified": [] }, { "id": "physical-intelligence-pi0-fast", "name": "π0-FAST", "level": "model", "family_id": "physical-intelligence-pi", "developer": [ "physical-intelligence" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": "autoregressive", "architecture_detail": "Autoregressive with FAST action tokenizer", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-02-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Autoregressive version of π0 released with open weights.", "verified_on": "2026-10-10", "sources": [ "https://www.pi.website/", "https://github.com/Physical-Intelligence/openpi" ], "unverified": [] }, { "id": "pika", "name": "Pika", "level": "family", "family_id": null, "developer": [ "pika" ], "model_class": "video generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Pika video models.", "verified_on": "2026-10-10", "sources": [ "https://pika.art/" ], "unverified": [] }, { "id": "pika-2-5", "name": "Pika 2.5", "level": "model", "family_id": "pika", "developer": [ "pika" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Pika's own video model; Pika now runs a multi-model platform reselling Seedance, MiniMax H3, Wan 3.0 and others.", "verified_on": "2026-10-10", "sources": [ "https://pika.art/", "https://pika.art/careers", "https://www.cliprise.app/news/pika-2-5-released" ], "unverified": [] }, { "id": "pika-music", "name": "Pika Music", "level": "model", "family_id": null, "developer": [ "pika" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Pika audio product (also Soundtrack, SFX).", "verified_on": "2026-10-10", "sources": [ "https://pika.art/" ], "unverified": [] }, { "id": "pika-speech", "name": "Pika Speech", "level": "model", "family_id": null, "developer": [ "pika" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Pika speech product; Indonesian not stated.", "verified_on": "2026-10-10", "sources": [ "https://pika.art/" ], "unverified": [] }, { "id": "pixverse-v6", "name": "PixVerse V6", "level": "model", "family_id": null, "developer": [ "pixverse" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-03", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "PixVerse video model.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/video/leaderboard/text-to-video", "https://www.tooljunction.io/ai-tools/pixverse/company" ], "unverified": [] }, { "id": "preferred-networks-plamo-3-610m-fin-instruct", "name": "PLaMo-3-610m-fin-instruct", "level": "model", "family_id": null, "developer": [ "preferred-networks" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 600000000, "params_active": null, "params_checkpoint": null, "params_note": "0.9B listed", "context_length": null, "release_date": "2026-10", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "laptop", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": "No Indonesian", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "PFN's small finance instruction model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/pfnet" ], "unverified": [ "licence" ] }, { "id": "preferred-networks-plamo-3-nict", "name": "PLaMo 3 NICT", "level": "family", "family_id": null, "developer": [ "preferred-networks", "nict" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-08-21", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Preferred Networks and NICT's PLaMo 3 models (31B, 8B, 2B base).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/pfnet" ], "unverified": [ "licence" ] }, { "id": "preferred-networks-plamo-3-nict-2b", "name": "PLaMo 3 NICT 2B", "level": "model", "family_id": "preferred-networks-plamo-3-nict", "developer": [ "preferred-networks", "nict" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 2000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-08-21", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "consumer GPU", "CPU server" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "PLaMo 3 NICT 2B base text model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/pfnet" ], "unverified": [ "licence" ] }, { "id": "preferred-networks-plamo-3-nict-31b", "name": "PLaMo 3 NICT 31B", "level": "model", "family_id": "preferred-networks-plamo-3-nict", "developer": [ "preferred-networks", "nict" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 31000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-08-21", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "consumer GPU", "CPU server" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "PLaMo 3 NICT 31B base text model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/pfnet" ], "unverified": [ "licence" ] }, { "id": "preferred-networks-plamo-3-nict-8b", "name": "PLaMo 3 NICT 8B", "level": "model", "family_id": "preferred-networks-plamo-3-nict", "developer": [ "preferred-networks", "nict" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-08-21", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "consumer GPU", "CPU server" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "PLaMo 3 NICT 8B base text model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/pfnet" ], "unverified": [ "licence" ] }, { "id": "prior-labs-realtabpfn-2-5", "name": "RealTabPFN-2.5", "level": "model", "family_id": "prior-labs-tabpfn", "developer": [ "prior-labs" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "dense transformer", "architecture_detail": "PFN continued-pretrained on curated real data", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download", "API" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Prior Labs best TFM on TabArena per Mindful Modeler; non-commercial weights, paid commercial licence.", "verified_on": "2026-10-10", "sources": [ "https://benchmarklist.com/benchmarks/tabarena_all/", "https://mindfulmodeler.substack.com/p/the-state-of-tabular-foundation-models" ], "unverified": [] }, { "id": "prior-labs-tabpfn", "name": "TabPFN", "level": "family", "family_id": null, "developer": [ "prior-labs", "ellis-tubingen" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "dense transformer", "architecture_detail": "In-context transformer (PFN)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Pretrained about 2 weeks on 8 GPUs on about 130M synthetic datasets; up to 10,000 samples and 500 features.", "context_length": null, "release_date": "2025-01-08", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Prior Labs tabular foundation model family.", "verified_on": "2026-10-10", "sources": [ "https://mindfulmodeler.substack.com/p/the-state-of-tabular-foundation-models", "https://docs.priorlabs.ai/changelog.md", "https://www.nature.com/articles/s41586-024-08328-6" ], "unverified": [] }, { "id": "prior-labs-tabpfn-2-6", "name": "TabPFN-2.6", "level": "model", "family_id": "prior-labs-tabpfn", "developer": [ "prior-labs" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "dense transformer", "architecture_detail": "In-context transformer (PFN)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-03-24", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "tabpfn-2-6-license-v1-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": "Median train about 5.5 s; tabpfn package Apache-2.0, 9.1.0 on 2 Oct 2026", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Prior Labs tabular foundation model; best TFM default on TabArena (rank 2).", "verified_on": "2026-10-10", "sources": [ "https://benchmarklist.com/benchmarks/tabarena_all/", "https://mindfulmodeler.substack.com/p/the-state-of-tabular-foundation-models", "https://huggingface.co/api/models/Prior-Labs/tabpfn_2_6", "https://pypi.org/pypi/tabpfn/json" ], "unverified": [] }, { "id": "prior-labs-tabpfn-3", "name": "TabPFN-3", "level": "model", "family_id": "prior-labs-tabpfn", "developer": [ "prior-labs" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": "Up to 1M rows, 200 features, 160 classes; Thinking mode", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Prior Labs TabPFN-3, available via API; open-weight status unverified.", "verified_on": "2026-10-10", "sources": [ "https://docs.priorlabs.ai/changelog.md" ], "unverified": [ "openness" ] }, { "id": "prior-labs-tabpfn-3-5", "name": "TabPFN-3.5", "level": "model", "family_id": "prior-labs-tabpfn", "developer": [ "prior-labs" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "dense transformer", "architecture_detail": "Single multitask checkpoint; 3.5-Fast in alpha", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-09", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "Up to 20,000 features, 1M rows", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Latest Prior Labs TabPFN, not yet benchmarked.", "verified_on": "2026-10-10", "sources": [ "https://docs.priorlabs.ai/changelog.md" ], "unverified": [ "licence" ] }, { "id": "prior-labs-tabpfn-ts", "name": "TabPFN-TS", "level": "model", "family_id": null, "developer": [ "prior-labs" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "PriorLabs zero-shot forecasting entry on GIFT-Eval.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/spaces/Salesforce/GIFT-Eval" ], "unverified": [] }, { "id": "prior-labs-tabpfn-v1", "name": "TabPFN v1", "level": "model", "family_id": "prior-labs-tabpfn", "developer": [ "prior-labs" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "TabPFN v1 (2022).", "verified_on": "2026-10-10", "sources": [ "https://mindfulmodeler.substack.com/p/the-state-of-tabular-foundation-models" ], "unverified": [] }, { "id": "prior-labs-tabpfn-v2", "name": "TabPFN v2", "level": "model", "family_id": "prior-labs-tabpfn", "developer": [ "prior-labs" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025", "release_date_basis": "notes", "weights_date": "2025-01-02", "openness": "open weights restricted", "licence": "priorlabs-1-1", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "TabPFN v2 (2025).", "verified_on": "2026-10-10", "sources": [ "https://mindfulmodeler.substack.com/p/the-state-of-tabular-foundation-models", "https://huggingface.co/api/models/Prior-Labs/TabPFN-v2-clf" ], "unverified": [] }, { "id": "pruna-ai-p-image", "name": "P-Image (P-Image 1.0)", "level": "model", "family_id": null, "developer": [ "pruna-ai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-12", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Pruna AI text-to-image model aimed at sub-second, low-cost generation with prompt adherence and text rendering.", "verified_on": "2026-10-11", "sources": [ "https://www.pruna.ai/performance-models-rayan/p-image", "https://artificialanalysis.ai/image/leaderboard/text-to-image" ], "unverified": [ "openness" ] }, { "id": "pyannoteai-pyannote-community-1", "name": "pyannote Community-1", "level": "model", "family_id": null, "developer": [ "pyannoteai" ], "model_class": "speaker", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open source", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open speaker diarization model; code MIT, weights CC-BY-4.0.", "verified_on": "2026-10-10", "sources": [ "https://www.pyannote.ai/blog/top-7-speaker-diarization-apis-and-models-in-2026" ], "unverified": [] }, { "id": "pyannoteai-pyannote-precision-2", "name": "pyannote Precision-2", "level": "model", "family_id": null, "developer": [ "pyannoteai" ], "model_class": "speaker", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "API or on-prem, from EUR 0.096/hr", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Commercial diarization model; vendor claims lowest DER in all ten benchmark domains, no figure given.", "verified_on": "2026-10-10", "sources": [ "https://www.pyannote.ai/blog/top-7-speaker-diarization-apis-and-models-in-2026" ], "unverified": [] }, { "id": "pythia-1b", "name": "Pythia-1B", "level": "model", "family_id": null, "developer": [], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": 1000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Model used as the academic pretraining cost anchor: about 41 days naive, 18 days optimised on 4 A100.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2410.23261v1" ], "unverified": [] }, { "id": "qure-ai-qxr", "name": "qXR", "level": "model", "family_id": null, "developer": [ "qure-ai" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": "CNN computer-aided detection software as a medical device", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Chest X-ray AI for TB screening and other findings. Qure.ai reports use in 105+ countries at 5,500+ sites and shows FDA and CE badges; piloted by Kemenkes at RS Fatmawati and RSPON in 2025.", "verified_on": "2026-10-10", "sources": [ "https://www.qure.ai/", "https://www.who.int/publications/i/item/9789240022676", "https://www.metrotvnews.com/read/K5nC7m81-kemenkes-manfaatkan-kecerdasan-buatan-untuk-deteksi-tuberkulosis" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "qwen2-5-vl", "name": "Qwen2.5-VL", "level": "model", "family_id": null, "developer": [], "model_class": "multimodal LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "VLM whose code UnifoLM-VLA inherits.", "verified_on": "2026-10-10", "sources": [ "https://raw.githubusercontent.com/unitreerobotics/unifolm-vla/main/README.md" ], "unverified": [] }, { "id": "randomforest", "name": "RandomForest", "level": "model", "family_id": null, "developer": [], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Baseline; TabArena Elo anchored so RandomForest (default) = 1000.", "verified_on": "2026-10-10", "sources": [ "https://benchmarklist.com/benchmarks/tabarena_all/" ], "unverified": [] }, { "id": "realmlp", "name": "RealMLP", "level": "model", "family_id": null, "developer": [ "inria" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-07-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Deep tabular (non-FM) model; best of its kind on TabArena.", "verified_on": "2026-10-10", "sources": [ "https://benchmarklist.com/benchmarks/tabarena_all/", "https://arxiv.org/html/2407.04491", "https://arxiv.org/abs/2407.04491" ], "unverified": [] }, { "id": "realtime-tts-2", "name": "Realtime TTS-2", "level": "model", "family_id": null, "developer": [], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-08", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "TTS model on the Artificial Analysis TTS arena; developer not shown.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/text-to-speech/leaderboard" ], "unverified": [] }, { "id": "recraft", "name": "Recraft", "level": "family", "family_id": null, "developer": [ "recraft" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Recraft image models, four generations (V2, V3, V4, V4.1).", "verified_on": "2026-10-10", "sources": [ "https://www.recraft.ai/about", "https://www.recraft.ai/docs/recraft-models/introduction" ], "unverified": [] }, { "id": "recraft-v2", "name": "Recraft V2", "level": "model", "family_id": "recraft", "developer": [ "recraft" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier Recraft generation.", "verified_on": "2026-10-10", "sources": [ "https://www.recraft.ai/docs/recraft-models/introduction" ], "unverified": [] }, { "id": "recraft-v3", "name": "Recraft V3", "level": "model", "family_id": "recraft", "developer": [ "recraft" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier Recraft generation.", "verified_on": "2026-10-10", "sources": [ "https://www.recraft.ai/docs/recraft-models/introduction" ], "unverified": [] }, { "id": "recraft-v4", "name": "Recraft V4", "level": "model", "family_id": "recraft", "developer": [ "recraft" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Recraft V4; V4 Styles family added to API 26 Aug 2026 (secondary).", "verified_on": "2026-10-10", "sources": [ "https://www.recraft.ai/docs/recraft-models/introduction", "https://usagepricing.com/blueprint/activity/recraft-2026-08-26-launch" ], "unverified": [] }, { "id": "recraft-v4-1", "name": "Recraft V4.1", "level": "model", "family_id": "recraft", "developer": [ "recraft" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app", "API" ], "deployment_tiers": [], "footprint_note": "Cloud only; Studio, API, Figma, Framer, Google Docs/Slides add-ons, MCP", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Latest Recraft image model, with vector and SVG output and text placement control.", "verified_on": "2026-10-10", "sources": [ "https://www.recraft.ai/about", "https://en.wikipedia.org/wiki/Recraft", "https://www.recraft.ai/docs/recraft-models/introduction" ], "unverified": [] }, { "id": "rednote-dots-ocr", "name": "dots.ocr", "level": "model", "family_id": null, "developer": [ "rednote" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "autoregressive", "architecture_detail": null, "params_total": 3000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-07-30", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "rednote document-parsing VLM.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2606.03264", "https://github.com/rednote-hilab/dots.ocr" ], "unverified": [] }, { "id": "rednote-firered-asr", "name": "FireRedASR", "level": "model", "family_id": null, "developer": [ "rednote" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ASR model from Xiaohongshu, named but not verified further.", "verified_on": "2026-10-10", "sources": [ "https://select.keenable.ai/r/speech-recognition-accuracy-leaders-2022-2026/JbURQjlVooLQ6J10NkbRVw" ], "unverified": [] }, { "id": "reka-ai-reka-flash-3", "name": "Reka Flash 3", "level": "model", "family_id": null, "developer": [ "reka-ai" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "Trained from scratch; released in Llama-compatible format", "params_total": 21000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 32000, "release_date": "2025-03-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": "39GB at fp16, about 11GB with 4-bit quantization", "runtimes": [], "languages_note": "Primarily English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Reka 21B general-purpose reasoning model released as an open research preview under Apache 2.0.", "verified_on": "2026-10-11", "sources": [ "https://reka.ai/news/introducing-reka-flash", "https://huggingface.co/RekaAI/reka-flash-3" ], "unverified": [ "deployment_tiers" ] }, { "id": "resemble-chatterbox", "name": "Resemble Chatterbox", "level": "model", "family_id": null, "developer": [ "resemble-ai" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open TTS model trending on Hugging Face (2025 to 2026).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/models?pipeline_tag=text-to-speech&sort=trending", "https://huggingface.co/ResembleAI/chatterbox" ], "unverified": [] }, { "id": "resnet50", "name": "ResNet50 (ImageNet)", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "ImageNet backbone used for transfer learning; developer not stated in these sections.", "verified_on": "2026-10-10", "sources": [ "https://shdl.mmu.edu.my/15767/" ], "unverified": [] }, { "id": "reve-2-0", "name": "Reve 2.0", "level": "model", "family_id": null, "developer": [ "reve" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier Reve image model.", "verified_on": "2026-10-10", "sources": [ "https://arena.ai/leaderboard/text-to-image" ], "unverified": [] }, { "id": "reve-2-1", "name": "Reve 2.1", "level": "model", "family_id": null, "developer": [ "reve" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Reve proprietary image model.", "verified_on": "2026-10-10", "sources": [ "https://arena.ai/leaderboard/text-to-image", "https://reve.com/about" ], "unverified": [] }, { "id": "rnnoise", "name": "RNNoise", "level": "model", "family_id": null, "developer": [], "model_class": "TinyML", "modalities_in": [ "text" ], "modalities_out": [ "labels and boxes" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "microcontroller", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Model listed in Arm ML-Zoo, optimised for Arm IP (Cortex-A, Cortex-M, Mali, Ethos-U) in TensorFlow Lite format.", "verified_on": "2026-10-10", "sources": [ "https://github.com/ARM-software/ML-zoo" ], "unverified": [] }, { "id": "roboflow-rf-detr", "name": "RF-DETR", "level": "family", "family_id": null, "developer": [ "roboflow" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image", "text" ], "architecture": "dense transformer", "architecture_detail": "Real-time DETR; LW-DETR base with DINOv2 backbone replacing CAEv2; neural architecture search", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "N 30.5M, S 32.1M, M 33.7M, 2XL 126.9M", "context_length": null, "release_date": "2025-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU", "embedded GPU" ], "footprint_note": "N 2.3 ms, 2XL 17.2 ms latency", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Roboflow real-time detection transformer family (N/S/M/L/XL/2XL); ICLR 2026 poster. Segmentation support per Roboflow is unverified.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2511.09554v2", "https://iclr.cc/virtual/2026/poster/10007257" ], "unverified": [ "licence" ] }, { "id": "roboflow-rf-detr-2xl", "name": "RF-DETR 2XL", "level": "variant", "family_id": "roboflow-rf-detr", "developer": [ "roboflow" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image", "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 126900000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU", "embedded GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "RF-DETR size variant 2XL.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2511.09554v2" ], "unverified": [ "licence" ] }, { "id": "roboflow-rf-detr-l", "name": "RF-DETR L", "level": "variant", "family_id": "roboflow-rf-detr", "developer": [ "roboflow" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image", "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU", "embedded GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "RF-DETR size variant L.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2511.09554v2" ], "unverified": [ "licence" ] }, { "id": "roboflow-rf-detr-m", "name": "RF-DETR M", "level": "variant", "family_id": "roboflow-rf-detr", "developer": [ "roboflow" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image", "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 33700000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU", "embedded GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "RF-DETR size variant M.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2511.09554v2", "https://github.com/roboflow/rf-detr" ], "unverified": [] }, { "id": "roboflow-rf-detr-n", "name": "RF-DETR N", "level": "variant", "family_id": "roboflow-rf-detr", "developer": [ "roboflow" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image", "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 30500000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU", "embedded GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "RF-DETR size variant N.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2511.09554v2", "https://github.com/roboflow/rf-detr" ], "unverified": [] }, { "id": "roboflow-rf-detr-s", "name": "RF-DETR S", "level": "variant", "family_id": "roboflow-rf-detr", "developer": [ "roboflow" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image", "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 32100000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU", "embedded GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "RF-DETR size variant S.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2511.09554v2", "https://github.com/roboflow/rf-detr" ], "unverified": [] }, { "id": "roboflow-rf-detr-xl", "name": "RF-DETR XL", "level": "variant", "family_id": "roboflow-rf-detr", "developer": [ "roboflow" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image", "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU", "embedded GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "RF-DETR size variant XL.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2511.09554v2" ], "unverified": [ "licence" ] }, { "id": "runway-aleph-2-0", "name": "Aleph 2.0", "level": "model", "family_id": null, "developer": [ "runway" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Runway video model.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Runway_(company)" ], "unverified": [] }, { "id": "runway-gen", "name": "Runway Gen", "level": "family", "family_id": null, "developer": [ "runway" ], "model_class": "video generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Runway Gen video model line.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Runway_(company)" ], "unverified": [] }, { "id": "runway-gen-3", "name": "Gen-3", "level": "model", "family_id": "runway-gen", "developer": [ "runway" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Runway video model scored in the WorldScore paper.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2504.00983v2" ], "unverified": [] }, { "id": "runway-gen-4", "name": "Gen-4", "level": "model", "family_id": "runway-gen", "developer": [ "runway" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-04", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Runway video model, Apr 2025.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Runway_(company)" ], "unverified": [] }, { "id": "runway-gen-4-5", "name": "Gen-4.5", "level": "model", "family_id": "runway-gen", "developer": [ "runway" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Runway video model; base of GWM-1. Runway has a custom-model deal with Lionsgate and AMC Networks.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Runway_(company)", "https://datanorth.ai/news/black-forest-labs-releases-flux-3", "https://runway.com/research/introducing-runway-gwm-1" ], "unverified": [] }, { "id": "runway-gwm-1", "name": "GWM-1", "level": "family", "family_id": null, "developer": [ "runway" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Runway General World Model 1 (Worlds, Avatars/Characters, Robotics).", "verified_on": "2026-10-10", "sources": [ "https://runway.com/research/introducing-runway-gwm-1" ], "unverified": [] }, { "id": "runway-gwm-1-characters", "name": "GWM-1 Characters", "level": "model", "family_id": "runway-gwm-1", "developer": [ "runway" ], "model_class": "world model", "modalities_in": [ "video", "action", "audio" ], "modalities_out": [ "video" ], "architecture": "autoregressive", "architecture_detail": "Autoregressive, built on Gen-4.5; 720p, up to 2 min; conditioned on camera pose, robot commands, audio", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-12-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "consumer app" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Runway avatar/character world model via API and web. One of three separate post-trained GWM-1 models; unification planned.", "verified_on": "2026-10-10", "sources": [ "https://runway.com/research/introducing-runway-gwm-1" ], "unverified": [] }, { "id": "runway-gwm-1-robotics", "name": "GWM-1 Robotics", "level": "model", "family_id": "runway-gwm-1", "developer": [ "runway" ], "model_class": "world model", "modalities_in": [ "video", "action", "audio" ], "modalities_out": [ "video" ], "architecture": "autoregressive", "architecture_detail": "Autoregressive, built on Gen-4.5; 720p, up to 2 min; conditioned on camera pose, robot commands, audio", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-12-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "gated preview" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Runway robotics world model with a Python SDK by request. One of three separate post-trained GWM-1 models; unification planned.", "verified_on": "2026-10-10", "sources": [ "https://runway.com/research/introducing-runway-gwm-1" ], "unverified": [] }, { "id": "runway-gwm-1-worlds", "name": "GWM-1 Worlds", "level": "model", "family_id": "runway-gwm-1", "developer": [ "runway" ], "model_class": "world model", "modalities_in": [ "video", "action", "audio" ], "modalities_out": [ "video" ], "architecture": "autoregressive", "architecture_detail": "Autoregressive, built on Gen-4.5; 720p, up to 2 min; conditioned on camera pose, robot commands, audio", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-12-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "gated preview" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Runway world model, early access. One of three separate post-trained GWM-1 models; unification planned.", "verified_on": "2026-10-10", "sources": [ "https://runway.com/research/introducing-runway-gwm-1" ], "unverified": [] }, { "id": "sakana-ai-fugu", "name": "Fugu", "level": "model", "family_id": null, "developer": [ "sakana-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-06-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multi-agent orchestration model that routes across other LLMs.", "verified_on": "2026-10-10", "sources": [ "https://sakana.ai/blog/" ], "unverified": [] }, { "id": "sakana-ai-fugu-max", "name": "Fugu Max", "level": "model", "family_id": null, "developer": [ "sakana-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-09-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://sakana.ai/blog/" ], "unverified": [] }, { "id": "sakana-ai-fugu-ultra-v2", "name": "Fugu Ultra v2", "level": "model", "family_id": null, "developer": [ "sakana-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-09-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://sakana.ai/blog/" ], "unverified": [] }, { "id": "sakana-ai-namazu", "name": "Namazu", "level": "family", "family_id": null, "developer": [ "sakana-ai" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-03-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": "Alpha 2026-03-24; Namazu API 2026-08-03.", "runtimes": [], "languages_note": "Japanese-specialised", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Sakana AI Japanese-specialised series that adapts open foundation models.", "verified_on": "2026-10-10", "sources": [ "https://sakana.ai/blog/" ], "unverified": [] }, { "id": "salesforce-moirai-1-0", "name": "Moirai 1.0", "level": "model", "family_id": null, "developer": [ "salesforce" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "dense transformer", "architecture_detail": "Masked encoder", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Predecessor of Moirai 2.0.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2511.11698", "https://www.salesforce.com/blog/moirai-2-0/" ], "unverified": [] }, { "id": "salesforce-moirai-2-0", "name": "Moirai 2.0", "level": "family", "family_id": null, "developer": [ "salesforce" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "dense transformer", "architecture_detail": "Decoder-only, single patch size, 9 quantile outputs, multi-token prediction", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "small 11.4M, base 87.1M, large 305M; only small released on HF", "context_length": null, "release_date": "2025-08-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server", "embedded GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Salesforce AI Research time-series foundation model; report revised 3 Feb 2026. CC BY-NC per secondary source; not stated in paper.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2511.11698", "https://www.beri.net/article/timesfm-vs-chronos-vs-moirai-time-series-foundation-model-licence-bigquery-sagemaker-cost-2026", "https://huggingface.co/spaces/Salesforce/GIFT-Eval", "https://www.salesforce.com/blog/moirai-2-0/", "https://arxiv.org/html/2511.11698v1", "https://huggingface.co/Salesforce/moirai-2.0-R-small", "https://arxiv.org/abs/2511.11698" ], "unverified": [] }, { "id": "salesforce-moirai-2-0-base", "name": "Moirai 2.0 base", "level": "variant", "family_id": "salesforce-moirai-2-0", "developer": [ "salesforce" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 87100000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "internal only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Moirai 2.0 base (not released on HF).", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2511.11698" ], "unverified": [] }, { "id": "salesforce-moirai-2-0-large", "name": "Moirai 2.0 large", "level": "variant", "family_id": "salesforce-moirai-2-0", "developer": [ "salesforce" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 305000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "internal only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Moirai 2.0 large (not released on HF).", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2511.11698" ], "unverified": [] }, { "id": "salesforce-moirai-2-0-r-small", "name": "Moirai 2.0 R small", "level": "variant", "family_id": null, "developer": [ "salesforce" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": null, "architecture_detail": null, "params_total": 11400000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Small Moirai 2.0 checkpoint, CC-BY-NC-4.0; code in uni2ts.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Salesforce/moirai-2.0-R-small" ], "unverified": [] }, { "id": "salesforce-moirai-2-0-small", "name": "Moirai 2.0 small", "level": "variant", "family_id": "salesforce-moirai-2-0", "developer": [ "salesforce" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 11400000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-11-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Moirai 2.0 small (the only size released on HF).", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2511.11698", "https://arxiv.org/abs/2511.11698" ], "unverified": [] }, { "id": "sam-vit-b", "name": "SAM ViT-B", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Segment Anything checkpoint from which MedSAM was initialized; developer not stated in these sections.", "verified_on": "2026-10-10", "sources": [ "https://github.com/bowang-lab/MedSAM" ], "unverified": [] }, { "id": "sand-ai-magi-1", "name": "MAGI-1", "level": "model", "family_id": null, "developer": [ "sand-ai" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-04", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Sand AI's openly licensed video model, Apr 2025.", "verified_on": "2026-10-10", "sources": [ "https://techcrunch.com/?p=2997035" ], "unverified": [ "openness" ] }, { "id": "sand-ai-magi-2-preview", "name": "MAGI-2 Preview", "level": "model", "family_id": null, "developer": [ "sand-ai" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-09", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": "Sand AI video model preview (0912).", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/video/leaderboard/text-to-video", "https://techcrunch.com/?p=2997035" ], "unverified": [] }, { "id": "sap-contexttab", "name": "ContextTab", "level": "model", "family_id": null, "developer": [ "sap" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": "Semantics-aware ICL", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "SAP tabular model; whether it is the same as SAP-RPT-OSS is unverified.", "verified_on": "2026-10-10", "sources": [ "https://mindfulmodeler.substack.com/p/the-state-of-tabular-foundation-models" ], "unverified": [] }, { "id": "sap-rpt-1-oss", "name": "SAP-RPT-OSS (sap-rpt-1-oss)", "level": "model", "family_id": null, "developer": [ "sap" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-06-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "SAP open tabular model on TabArena (Elo 1284, rank 28).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/api/models/SAP/sap-rpt-1-oss", "https://benchmarklist.com/benchmarks/tabarena_all/" ], "unverified": [] }, { "id": "sapiens-ai-agnes-video-2-5", "name": "Agnes-Video-2.5", "level": "model", "family_id": null, "developer": [ "sapiens-ai" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-08", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Sapiens AI video model on Artificial Analysis.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/video/leaderboard/text-to-video" ], "unverified": [] }, { "id": "sarvam-ai-sarvam-105b", "name": "Sarvam 105B", "level": "model", "family_id": null, "developer": [ "sarvam-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "MoE with MLA-style attention (128 experts, top-8 + 1 shared)", "params_total": 105000000000, "params_active": 10300000000, "params_checkpoint": null, "params_note": "Context 128K via YaRN", "context_length": 128000, "release_date": "2026-03", "release_date_basis": "notes", "weights_date": "2026-03-06", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": "Single GPU at FP8; GGUF variant", "runtimes": [], "languages_note": "Indic focus", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open-sourced 2026-03-06.", "verified_on": "2026-10-10", "sources": [ "https://www.sarvam.ai/blogs", "https://artificialanalysis.ai/leaderboards/models", "https://huggingface.co/sarvamai", "https://huggingface.co/sarvamai/sarvam-105b", "https://huggingface.co/sarvamai/sarvam-105b/raw/main/README.md", "https://huggingface.co/api/models/sarvamai/sarvam-105b" ], "unverified": [] }, { "id": "sarvam-ai-sarvam-30b", "name": "Sarvam 30B", "level": "model", "family_id": null, "developer": [ "sarvam-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 30000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-03", "release_date_basis": "registry date", "weights_date": "2026-03-06", "openness": "unknown", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [ "consumer GPU" ], "footprint_note": "GGUF", "runtimes": [], "languages_note": "Indic focus", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open-sourced 2026-03-06.", "verified_on": "2026-10-10", "sources": [ "https://www.sarvam.ai/blogs", "https://huggingface.co/sarvamai", "https://huggingface.co/sarvamai/sarvam-105b", "https://huggingface.co/sarvamai/sarvam-30b/raw/main/README.md", "https://huggingface.co/api/models/sarvamai/sarvam-30b", "https://huggingface.co/sarvamai/sarvam-30b" ], "unverified": [] }, { "id": "sarvam-ai-sarvam-vision-2-1", "name": "Sarvam Vision 2.1", "level": "model", "family_id": null, "developer": [ "sarvam-ai" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-09-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://www.sarvam.ai/blogs" ], "unverified": [] }, { "id": "screenpoint-medical-transpara", "name": "Transpara", "level": "model", "family_id": null, "developer": [ "screenpoint-medical" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": "CNN detection and triage software as a medical device", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Mammography cancer detection software. Its use in the Swedish MASAI trial could not be confirmed because the source failed to load.", "verified_on": "2026-10-10", "sources": [ "https://www.thelancet.com/journals/lanonc/article/PIIS1470-2045(23)00298-X/fulltext" ], "unverified": [] }, { "id": "seasonal-naive", "name": "Seasonal_Naive", "level": "model", "family_id": null, "developer": [], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Baseline that GIFT-Eval scores are normalised against.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/spaces/Salesforce/GIFT-Eval" ], "unverified": [] }, { "id": "sensenova-u1-a3b-mot", "name": "SenseNova-U1-A3B-MoT", "level": "model", "family_id": null, "developer": [], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "MoE", "architecture_detail": "Mixture-of-Transformers with stream-wise mixture of experts: 128 understanding experts and 32 generation experts, top-8 routing per stream", "params_total": null, "params_active": 3000000000, "params_checkpoint": null, "params_note": "30.0B understanding experts, 8.2B generation experts, about 3B active per the paper; GenEval table row 8BA3B", "context_length": null, "release_date": "2026-05-12", "release_date_basis": "inferred from page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "SenseNova-U1 variant that unifies multimodal understanding and generation in a mixture-of-experts design with about 3B active parameters.", "verified_on": "2026-10-11", "sources": [ "https://arxiv.org/abs/2605.12500" ], "unverified": [] }, { "id": "sentence-transformers-all-minilm-l6-v2", "name": "all-MiniLM-L6-v2", "level": "model", "family_id": null, "developer": [ "sentence-transformers" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "MiniLM-L6-H384 (BERT type) fine-tuned with a contrastive objective on about 1.17B sentence pairs; 384-dimension embeddings", "params_total": 22700000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 256, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server", "laptop" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Sentence-transformers model that maps sentences and paragraphs to 384-dimension vectors for clustering and semantic search.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/sentence-transformers/all-MiniLM-L6-v2" ], "unverified": [ "deployment_tiers" ] }, { "id": "sentence-transformers-paraphrase-multilingual-minilm-l12-v2", "name": "paraphrase-multilingual-MiniLM-L12-v2", "level": "model", "family_id": null, "developer": [ "sentence-transformers" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [], "architecture": "other", "architecture_detail": "BertModel transformer with mean pooling to a 384-dimensional vector", "params_total": 117654272, "params_active": null, "params_checkpoint": null, "params_note": "Hugging Face API Safetensors count", "context_length": 128, "release_date": "2022-03-02", "release_date_basis": "registry date", "weights_date": "2022-03-02", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Card language tags list about 50 languages including Indonesian, Malay, Thai and Vietnamese", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "A multilingual sentence-transformers model mapping sentences and paragraphs to a 384-dimensional vector for clustering and semantic search.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2", "https://huggingface.co/api/models/sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "sentence-transformers-paraphrase-multilingual-mpnet-base-v2", "name": "paraphrase-multilingual-mpnet-base-v2", "level": "model", "family_id": null, "developer": [ "sentence-transformers" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [], "architecture": "other", "architecture_detail": "XLMRobertaModel transformer with mean pooling to a 768-dimensional vector", "params_total": 278044162, "params_active": null, "params_checkpoint": null, "params_note": "Hugging Face API Safetensors count", "context_length": 128, "release_date": "2022-03-02", "release_date_basis": "registry date", "weights_date": "2022-03-02", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Card language tags list about 50 languages including Indonesian, Malay, Thai and Vietnamese", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "A multilingual sentence-transformers model mapping sentences and paragraphs to a 768-dimensional vector for clustering and semantic search.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/sentence-transformers/paraphrase-multilingual-mpnet-base-v2", "https://huggingface.co/api/models/sentence-transformers/paraphrase-multilingual-mpnet-base-v2" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "seoul-national-university-sevennet-omni-i12", "name": "SevenNet-Omni-i12", "level": "model", "family_id": null, "developer": [ "seoul-national-university" ], "model_class": "materials", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "GNN", "architecture_detail": "Equivariant GNN interatomic potential (MLIP)", "params_total": 54900000, "params_active": null, "params_checkpoint": null, "params_note": "Training data 243M COSMOSDataset", "context_length": null, "release_date": "2026-01-12", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Machine-learned interatomic potential on Matbench Discovery (rank 6 by CPS, date added 2026-01-12).", "verified_on": "2026-10-10", "sources": [ "https://matbench-discovery.materialsproject.org/" ], "unverified": [ "developer" ] }, { "id": "sesame-csm-1b", "name": "CSM-1B", "level": "model", "family_id": null, "developer": [ "sesame" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "dense transformer", "architecture_detail": "Llama backbone plus audio decoder emitting Mimi codes", "params_total": 1000000000, "params_active": null, "params_checkpoint": null, "params_note": "1B (metadata 2B)", "context_length": null, "release_date": "2025-03-13", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": "English; weak non-English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Sesame's open conversational speech model; Sesame also runs voice agents in iOS/Android apps.", "verified_on": "2026-10-10", "sources": [ "https://www.sesame.com/", "https://huggingface.co/sesame/csm-1b", "https://www.sesame.com/about" ], "unverified": [ "deployment_tiers" ] }, { "id": "sesr", "name": "SESR", "level": "model", "family_id": null, "developer": [], "model_class": "TinyML", "modalities_in": [ "text" ], "modalities_out": [ "labels and boxes" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "microcontroller", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Model listed in Arm ML-Zoo, optimised for Arm IP (Cortex-A, Cortex-M, Mali, Ethos-U) in TensorFlow Lite format.", "verified_on": "2026-10-10", "sources": [ "https://github.com/ARM-software/ML-zoo" ], "unverified": [] }, { "id": "shanghai-ai-lab-fengwu", "name": "FengWu", "level": "model", "family_id": null, "developer": [ "shanghai-ai-lab" ], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Shanghai AI Laboratory weather model (Techwind for industry); reported beating GraphCast on about 80% of variables; distributed as fengwu_v1.onnx and fengwu_v2.onnx.", "verified_on": "2026-10-10", "sources": [ "https://news.tuoitre.vn/china-bets-on-ai-weather-forecasting-as-extreme-weather-intensifies-103260812213327362.htm", "https://infoplaza.com/en/blog/ai-weather-forecasting-second-generation-ai-weather-models", "https://github.com/OpenEarthLab/FengWu" ], "unverified": [ "architecture", "licence" ] }, { "id": "shanghai-ai-lab-fengwu-4dvar", "name": "FengWu-4DVar", "level": "variant", "family_id": "shanghai-ai-lab-fengwu", "developer": [ "shanghai-ai-lab" ], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "FengWu variant.", "verified_on": "2026-10-10", "sources": [ "https://news.tuoitre.vn/china-bets-on-ai-weather-forecasting-as-extreme-weather-intensifies-103260812213327362.htm" ], "unverified": [] }, { "id": "shanghai-ai-lab-fengwu-ghr", "name": "FengWu-GHR", "level": "variant", "family_id": "shanghai-ai-lab-fengwu", "developer": [ "shanghai-ai-lab" ], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "FengWu variant.", "verified_on": "2026-10-10", "sources": [ "https://news.tuoitre.vn/china-bets-on-ai-weather-forecasting-as-extreme-weather-intensifies-103260812213327362.htm" ], "unverified": [] }, { "id": "shanghai-ai-lab-intern-s1-pro", "name": "Intern-S1-Pro", "level": "model", "family_id": null, "developer": [ "shanghai-ai-lab" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 917000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-10", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "multi-GPU node" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Shanghai AI Lab vision-language model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/internlm" ], "unverified": [ "architecture", "licence" ] }, { "id": "shanghai-ai-lab-intern-s2-397b", "name": "Intern-S2-397B", "level": "model", "family_id": null, "developer": [ "shanghai-ai-lab" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 403000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-09", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "multi-GPU node" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Shanghai AI Lab vision-language model; the 397B size matches Qwen3.5-397B but lineage is not confirmed.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/internlm" ], "unverified": [ "architecture", "licence" ] }, { "id": "shanghai-ai-laboratory-atria-dawn-preview", "name": "Atria Dawn Preview", "level": "model", "family_id": null, "developer": [ "shanghai-ai-lab" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-09-11", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": "Shanghai AI Laboratory open-weights model top on BrowseComp per BenchLM.", "verified_on": "2026-10-10", "sources": [ "https://www.benchlm.ai/benchmarks/browsecomp", "https://huggingface.co/internlm/Atria-Dawn-Preview" ], "unverified": [] }, { "id": "shengshu-technology-vidu", "name": "Vidu", "level": "family", "family_id": null, "developer": [ "shengshu-technology" ], "model_class": "video generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ShengShu Vidu video line, launched Apr 2024 with Tsinghua University.", "verified_on": "2026-10-10", "sources": [ "https://www.cdomagazine.tech/aiml/chinas-vidu-emerges-as-sora-challenger-in-text-to-video-ai-race", "https://tirto.id/vidu-showcases-china-speed-in-advancing-ai-video-into-production-at-global-creativity-week-hqfJ", "https://www.vidu.com/" ], "unverified": [] }, { "id": "shengshu-technology-vidu-q1", "name": "Vidu Q1", "level": "model", "family_id": "shengshu-technology-vidu", "developer": [ "shengshu-technology" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-04-17", "release_date_basis": "registry date", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vidu Q1, on VBench and VBench-2.0.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/spaces/Vchitect/VBench_Leaderboard" ], "unverified": [] }, { "id": "shengshu-technology-vidu-q2", "name": "Vidu Q2", "level": "model", "family_id": "shengshu-technology-vidu", "developer": [ "shengshu-technology" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vidu Q2 (Reference-to-Video Pro).", "verified_on": "2026-10-10", "sources": [ "https://tirto.id/vidu-showcases-china-speed-in-advancing-ai-video-into-production-at-global-creativity-week-hqfJ" ], "unverified": [] }, { "id": "shengshu-technology-vidu-q3-pro", "name": "Vidu Q3 Pro", "level": "model", "family_id": "shengshu-technology-vidu", "developer": [ "shengshu-technology" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ShengShu video model; Q3 was the first Vidu with long-form native audio and video per the vendor.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/video/leaderboard/text-to-video", "https://tirto.id/vidu-showcases-china-speed-in-advancing-ai-video-into-production-at-global-creativity-week-hqfJ", "https://alphasignal.ai/news/shengshu-s-vidu-q4-preview-debuts-at-no-3-for-ai-video" ], "unverified": [] }, { "id": "shengshu-technology-vidu-q3-turbo", "name": "Vidu Q3 Turbo", "level": "model", "family_id": "shengshu-technology-vidu", "developer": [ "shengshu-technology" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Turbo version of Vidu Q3.", "verified_on": "2026-10-10", "sources": [ "https://alphasignal.ai/news/shengshu-s-vidu-q4-preview-debuts-at-no-3-for-ai-video" ], "unverified": [] }, { "id": "shengshu-technology-vidu-q4-preview", "name": "Vidu Q4 Preview", "level": "model", "family_id": "shengshu-technology-vidu", "developer": [ "shengshu-technology" ], "model_class": "video generation", "modalities_in": [ "image" ], "modalities_out": [ "video", "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-10-08", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": "ShengShu video model (image-to-video and reference-to-video only), 3 to 16 s at up to 4K with native audio, US$7.20 per minute at 1080p.", "verified_on": "2026-10-10", "sources": [ "https://alphasignal.ai/news/shengshu-s-vidu-q4-preview-debuts-at-no-3-for-ai-video", "https://www.vidu.com/" ], "unverified": [] }, { "id": "shengshu-technology-vidu-s", "name": "Vidu S", "level": "model", "family_id": "shengshu-technology-vidu", "developer": [ "shengshu-technology" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time Vidu model listed on the Vidu site.", "verified_on": "2026-10-10", "sources": [ "https://www.vidu.com/" ], "unverified": [] }, { "id": "siemens-industrial-foundation-model", "name": "Siemens Industrial Foundation Model (IFM)", "level": "model", "family_id": null, "developer": [ "siemens", "microsoft" ], "model_class": "multimodal LLM", "modalities_in": [ "3D", "image", "text" ], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": "Azure", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "announced", "summary": "Siemens model announced in development on 31 Mar 2025 to process 3D models, 2D drawings, industrial data and specs; architecture and size undisclosed. Release status at Oct 2026 unverified.", "verified_on": "2026-10-10", "sources": [ "https://press.siemens.com/global/en/node/7143", "https://www.arcweb.com/blog/global-industrial-ai-revolution-nvidias-european-investment-siemens-worldwide-vision" ], "unverified": [] }, { "id": "siglip-400m", "name": "SigLIP-400M", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Image encoder from which MedSigLIP is derived; developer not stated in these sections.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2507.05201v2" ], "unverified": [] }, { "id": "silero", "name": "Silero", "level": "model", "family_id": null, "developer": [], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Model entry listed on PyTorch Hub.", "verified_on": "2026-10-10", "sources": [ "https://pytorch.org/hub/" ], "unverified": [ "model_class" ] }, { "id": "simba-3-2", "name": "Simba 3.2", "level": "model", "family_id": null, "developer": [], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-07", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "TTS model on the Artificial Analysis TTS arena; developer not shown.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/text-to-speech/leaderboard" ], "unverified": [] }, { "id": "sk-telecom-a-x", "name": "A.X", "level": "family", "family_id": null, "developer": [ "sk-telecom" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "A.X is the SK Telecom family of foundation models.", "verified_on": "2026-10-11", "sources": [ "https://news.sktelecom.com/en/3204" ], "unverified": [] }, { "id": "sk-telecom-a-x-k2", "name": "A.X K2", "level": "model", "family_id": "sk-telecom-a-x", "developer": [ "sk-telecom" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "Decoder-only MoE, 61 layers, 256 routed experts plus 1 shared (8 active), Multi-head Latent Attention with Sparse Gated Attention", "params_total": 688000000000, "params_active": 33000000000, "params_checkpoint": null, "params_note": null, "context_length": 262144, "release_date": "2026-07-29", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English, Korean, Chinese, Japanese, Spanish", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "SK Telecom 688B foundation model trained from scratch, released on Hugging Face, with a focus on Korean knowledge and long-context reasoning.", "verified_on": "2026-10-11", "sources": [ "https://news.sktelecom.com/en/3204", "https://huggingface.co/skt/A.X-K2" ], "unverified": [] }, { "id": "skywork-ai-skyreels-v4", "name": "SkyReels V4", "level": "model", "family_id": null, "developer": [ "skywork-ai" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-03", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Skywork AI video model.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/video/leaderboard/text-to-video", "https://www.everydev.ai/developers/skyreels-v4" ], "unverified": [] }, { "id": "softbank-sarashina", "name": "Sarashina", "level": "model", "family_id": null, "developer": [ "softbank" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Not researched (search budget).", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/leaderboards/models" ], "unverified": [ "status" ] }, { "id": "spacexai-aurora", "name": "Aurora", "level": "model", "family_id": null, "developer": [ "spacexai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-12", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "xAI's own image model used in Grok from Dec 2024, after Grok used FLUX in Aug 2024.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Black_Forest_Labs" ], "unverified": [] }, { "id": "spacexai-grok", "name": "Grok", "level": "family", "family_id": null, "developer": [ "spacexai" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://docs.x.ai/docs/models" ], "unverified": [] }, { "id": "spacexai-grok-2-1212", "name": "Grok 2 (Dec 2024, grok-2-1212)", "level": "model", "family_id": "spacexai-grok", "developer": [ "spacexai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 131000, "release_date": "2024-12-12", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "December 2024 checkpoint of xAI Grok 2, non-reasoning; Artificial Analysis lists it as deprecated and as open weights at 270B parameters, which no developer page opened here confirms.", "verified_on": "2026-10-11", "sources": [ "https://artificialanalysis.ai/models/grok-2-1212" ], "unverified": [ "context_length" ] }, { "id": "spacexai-grok-3", "name": "Grok-3", "level": "model", "family_id": null, "developer": [ "spacexai" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-02", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "xAI model named by Epoch as the first model trained above 1e26 FLOP, in February 2025.", "verified_on": "2026-10-10", "sources": [ "https://epoch.ai/tools/model-counts" ], "unverified": [] }, { "id": "spacexai-grok-4", "name": "Grok 4", "level": "model", "family_id": "spacexai-grok", "developer": [ "spacexai" ], "model_class": "multimodal LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-07-09", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "xAI reasoning model whose RL stage xAI states ran at pretraining scale on its 200,000 GPU Colossus cluster.", "verified_on": "2026-10-10", "sources": [ "https://docs.x.ai/docs/models", "https://www.digitalapplied.com/blog/ai-model-releases-october-2026-tracker", "https://docs.github.com/en/copilot/reference/ai-models/supported-models", "https://x.ai/news/grok-4" ], "unverified": [] }, { "id": "spacexai-grok-4-1-fast", "name": "Grok 4.1 Fast", "level": "model", "family_id": "spacexai-grok", "developer": [ "spacexai" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 2000000, "release_date": "2025-11-19", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "xAI tool-calling model with a 2 million token context window, trained with reinforcement learning for agentic use such as customer support and finance, launched with the Agent Tools API.", "verified_on": "2026-10-11", "sources": [ "https://x.ai/news/grok-4-1-fast", "https://docs.x.ai/docs/release-notes" ], "unverified": [ "status" ] }, { "id": "spacexai-grok-4-20", "name": "Grok 4.20", "level": "model", "family_id": "spacexai-grok", "developer": [ "spacexai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Grok 4.20, listed on Arena Text as grok-4.20-beta1.", "verified_on": "2026-10-10", "sources": [ "https://docs.x.ai/docs/models", "https://www.digitalapplied.com/blog/ai-model-releases-october-2026-tracker", "https://docs.github.com/en/copilot/reference/ai-models/supported-models" ], "unverified": [] }, { "id": "spacexai-grok-4-3", "name": "Grok 4.3", "level": "model", "family_id": "spacexai-grok", "developer": [ "spacexai" ], "model_class": "reasoning", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": "2026-04-30", "release_date_basis": "registry date", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "Reasoning effort none, low, medium or high; higher rates above 200K context; API region us-east-1", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "SpaceXAI (xAI) reasoning model with text and image input, function calling and a one-million-token context window, served through the xAI API.", "verified_on": "2026-10-11", "sources": [ "https://docs.x.ai/docs/models/grok-4.3", "https://openrouter.ai/x-ai/grok-4.3" ], "unverified": [] }, { "id": "spacexai-grok-4-5", "name": "Grok 4.5", "level": "model", "family_id": "spacexai-grok", "developer": [ "spacexai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://docs.x.ai/docs/models", "https://www.digitalapplied.com/blog/ai-model-releases-october-2026-tracker", "https://docs.github.com/en/copilot/reference/ai-models/supported-models" ], "unverified": [] }, { "id": "spacexai-grok-4-6", "name": "Grok 4.6", "level": "model", "family_id": "spacexai-grok", "developer": [ "spacexai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-08-12", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://docs.x.ai/docs/models", "https://www.digitalapplied.com/blog/ai-model-releases-october-2026-tracker", "https://docs.github.com/en/copilot/reference/ai-models/supported-models" ], "unverified": [] }, { "id": "spacexai-grok-4-7", "name": "Grok 4.7", "level": "model", "family_id": "spacexai-grok", "developer": [ "spacexai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 500000, "release_date": "2026-09-21", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "Model ID grok-4.7. Regions not stated on xAI docs. Configurable reasoning.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "SpaceXAI's current Grok model; xAI recommends it for code (no code-specific model listed). GA in GitHub Copilot.", "verified_on": "2026-10-10", "sources": [ "https://docs.x.ai/docs/models", "https://artificialanalysis.ai/models/grok-4-7", "https://artificialanalysis.ai/leaderboards/models", "https://docs.x.ai/docs/models", "https://docs.github.com/en/copilot/reference/ai-models/supported-models" ], "unverified": [] }, { "id": "spacexai-grok-code-fast-1", "name": "grok-code-fast-1", "level": "model", "family_id": "spacexai-grok", "developer": [ "spacexai" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": "Not disclosed", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 256000, "release_date": "2025-08-28", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "xAI coding model; not on the current xAI model list.", "verified_on": "2026-10-10", "sources": [ "https://www.verdent.ai/guides/grok-for-coding-2026", "https://docs.x.ai/docs/models" ], "unverified": [ "status" ] }, { "id": "spacexai-grok-imagine", "name": "Grok Imagine", "level": "family", "family_id": null, "developer": [ "spacexai" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "xAI (SpaceXAI) image and video generation line used in Grok.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text", "https://arena.ai/leaderboard/text-to-image", "https://artificialanalysis.ai/video/leaderboard/text-to-video" ], "unverified": [] }, { "id": "spacexai-grok-imagine-image", "name": "Grok Imagine Image", "level": "model", "family_id": "spacexai-grok-imagine", "developer": [ "spacexai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier grok-imagine-image model.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text" ], "unverified": [] }, { "id": "spacexai-grok-imagine-image-2-0", "name": "Grok Imagine Image 2.0", "level": "model", "family_id": "spacexai-grok-imagine", "developer": [ "spacexai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-08", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "xAI media or voice sibling, outside the scope of the research notes.", "verified_on": "2026-10-10", "sources": [ "https://docs.x.ai/docs/models", "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text", "https://arena.ai/leaderboard/text-to-image", "https://en.wikipedia.org/wiki/Black_Forest_Labs", "https://en.wikipedia.org/wiki/SpaceXAI" ], "unverified": [] }, { "id": "spacexai-grok-imagine-video-1-5", "name": "Grok Imagine Video 1.5", "level": "model", "family_id": "spacexai-grok-imagine", "developer": [ "spacexai" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-05", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "xAI media or voice sibling, outside the scope of the research notes.", "verified_on": "2026-10-10", "sources": [ "https://docs.x.ai/docs/models", "https://artificialanalysis.ai/video/leaderboard/text-to-video", "https://arena.ai/leaderboard/text-to-video", "https://en.wikipedia.org/wiki/SpaceXAI" ], "unverified": [] }, { "id": "spacexai-grok-imagine-video-720p", "name": "Grok Imagine Video 720p", "level": "model", "family_id": "spacexai-grok-imagine", "developer": [ "spacexai" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier grok-imagine-video-720p model.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/video/leaderboard/text-to-video" ], "unverified": [] }, { "id": "spacexai-grok-voice", "name": "Grok Voice API", "level": "model", "family_id": null, "developer": [ "spacexai" ], "model_class": "speech synthesis", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "xAI media or voice sibling, outside the scope of the research notes.", "verified_on": "2026-10-10", "sources": [ "https://docs.x.ai/docs/models" ], "unverified": [] }, { "id": "spacexai-tts", "name": "SpaceXAI TTS", "level": "model", "family_id": null, "developer": [ "spacexai" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-03", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "TTS model listed on the Artificial Analysis TTS arena under the SpaceXAI name.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/text-to-speech/leaderboard" ], "unverified": [ "developer" ] }, { "id": "spark-4-0-2024-10-14", "name": "Spark 4.0-2024-10-14", "level": "model", "family_id": null, "developer": [], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Model top on CMMLU zero-shot.", "verified_on": "2026-10-10", "sources": [ "https://github.com/haonan-li/CMMLU" ], "unverified": [] }, { "id": "speechbrain-ecapa-tdnn", "name": "ECAPA-TDNN (SpeechBrain)", "level": "model", "family_id": null, "developer": [ "speechbrain" ], "model_class": "speaker", "modalities_in": [ "audio" ], "modalities_out": [], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Speaker verification embedding model.", "verified_on": "2026-10-10", "sources": [ "https://www.pyannote.ai/blog/top-7-speaker-diarization-apis-and-models-in-2026" ], "unverified": [] }, { "id": "ssd-mobilenet-v1", "name": "SSD MobileNet v1", "level": "model", "family_id": null, "developer": [], "model_class": "TinyML", "modalities_in": [ "text" ], "modalities_out": [ "labels and boxes" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "microcontroller", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Model listed in Arm ML-Zoo, optimised for Arm IP (Cortex-A, Cortex-M, Mali, Ethos-U) in TensorFlow Lite format.", "verified_on": "2026-10-10", "sources": [ "https://github.com/ARM-software/ML-zoo" ], "unverified": [] }, { "id": "stability-ai-sdxl", "name": "SDXL", "level": "model", "family_id": "stability-ai-stable-diffusion", "developer": [ "stability-ai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "SDXL base model, still 4.3M downloads with a large LoRA ecosystem.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/models?pipeline_tag=text-to-image&sort=trending", "https://en.wikipedia.org/wiki/Stability_AI" ], "unverified": [] }, { "id": "stability-ai-spar3d", "name": "SPAR3D", "level": "model", "family_id": null, "developer": [ "stability-ai" ], "model_class": "3D", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Stability AI image-to-3D model trending on Hugging Face.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/models?pipeline_tag=image-to-3d&sort=trending" ], "unverified": [ "licence" ] }, { "id": "stability-ai-stable-audio", "name": "Stable Audio", "level": "family", "family_id": null, "developer": [ "stability-ai" ], "model_class": "music and audio", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Stability AI audio generation line.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/stabilityai/stable-audio-3-medium" ], "unverified": [] }, { "id": "stability-ai-stable-audio-2-5", "name": "Stable Audio 2.5", "level": "model", "family_id": "stability-ai-stable-audio", "developer": [ "stability-ai" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier Stable Audio model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/stabilityai/stable-audio-3-medium" ], "unverified": [] }, { "id": "stability-ai-stable-audio-3-medium", "name": "Stable Audio 3 Medium", "level": "model", "family_id": "stability-ai-stable-audio", "developer": [ "stability-ai" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "diffusion", "architecture_detail": "Transformer latent diffusion, T5Gemma conditioning, adversarial post-training", "params_total": 2000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-05-18", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "stability-community", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": "English only", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Stability AI open-weight audio model (also Small Music and Small SFX), trained on licensed AudioSparx and CC Freesound data.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/stabilityai/stable-audio-3-medium", "https://huggingface.co/models?pipeline_tag=text-to-audio&sort=trending", "https://en.wikipedia.org/wiki/Stability_AI" ], "unverified": [ "deployment_tiers" ] }, { "id": "stability-ai-stable-audio-open-1-0", "name": "Stable Audio Open 1.0", "level": "model", "family_id": "stability-ai-stable-audio", "developer": [ "stability-ai" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier open Stable Audio model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/stabilityai/stable-audio-3-medium" ], "unverified": [] }, { "id": "stability-ai-stable-diffusion", "name": "Stable Diffusion", "level": "family", "family_id": null, "developer": [ "stability-ai" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Stability AI Stable Diffusion image models.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Stability_AI" ], "unverified": [] }, { "id": "stability-ai-stable-diffusion-3-5", "name": "Stable Diffusion 3.5", "level": "model", "family_id": "stability-ai-stable-diffusion", "developer": [ "stability-ai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": "MMDiT (unverified)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights restricted", "licence": "stability-community", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Stability AI open-weight image model. UK High Court found no copyright infringement in Getty v. Stability (Nov 2025).", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Stability_AI", "https://huggingface.co/models?pipeline_tag=text-to-image&sort=trending" ], "unverified": [ "architecture_detail", "licence", "deployment_tiers" ] }, { "id": "stability-ai-stable-fast-3d", "name": "Stable Fast 3D", "level": "model", "family_id": null, "developer": [ "stability-ai" ], "model_class": "3D", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Stability AI image-to-3D model trending on Hugging Face.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/models?pipeline_tag=image-to-3d&sort=trending", "https://en.wikipedia.org/wiki/Stability_AI" ], "unverified": [ "licence", "deployment_tiers" ] }, { "id": "stability-ai-triposr", "name": "TripoSR", "level": "model", "family_id": null, "developer": [ "stability-ai", "vast" ], "model_class": "3D", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Image-to-3D model co-developed by Stability AI and Tripo/VAST.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/models?pipeline_tag=image-to-3d&sort=trending", "https://investgame.net/news/matrix-partners-china-has-led-a-446m-series-b-and-b-into-vast-maker-of-tripo-ai-in-its-fourth-round-of-2026/" ], "unverified": [] }, { "id": "stableai-limix", "name": "LimiX", "level": "model", "family_id": null, "developer": [ "stableai" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": "Joint-distribution model including missingness", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "StableAI tabular foundation model.", "verified_on": "2026-10-10", "sources": [ "https://mindfulmodeler.substack.com/p/the-state-of-tabular-foundation-models" ], "unverified": [ "licence" ] }, { "id": "stanford-university-chexagent", "name": "CheXagent", "level": "model", "family_id": null, "developer": [ "stanford-university" ], "model_class": "multimodal LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Radiology report generation model attributed to Stanford; unverified in the notes.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/microsoft/maira-2", "https://arxiv.org/html/2507.05201v2" ], "unverified": [ "developer", "indonesian_measured", "indonesia_hosting" ] }, { "id": "stanford-university-openvla", "name": "OpenVLA", "level": "model", "family_id": "stanford-university-openvla-family", "developer": [ "stanford-university", "uc-berkeley", "toyota-research-institute" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": "dense transformer", "architecture_detail": "Llama 2 LLM plus fused DINOv2 and SigLIP vision encoder", "params_total": 7000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-06", "release_date_basis": "registry date", "weights_date": null, "openness": "open source", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open 7B VLA that beats RT-2-X by 16.5 points over 29 tasks.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2406.09246", "https://arxiv.org/abs/2502.19645" ], "unverified": [ "developer", "licence" ] }, { "id": "stanford-university-openvla-family", "name": "OpenVLA", "level": "family", "family_id": null, "developer": [ "stanford-university" ], "model_class": "robot policy", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open vision-language-action models from Stanford and partners.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2406.09246" ], "unverified": [] }, { "id": "stanford-university-openvla-oft", "name": "OpenVLA-OFT", "level": "model", "family_id": "stanford-university-openvla-family", "developer": [ "stanford-university" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": null, "architecture_detail": "Optimised fine-tuning recipe: parallel decoding, action chunking, continuous actions", "params_total": 7000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-02-27", "release_date_basis": "registry date", "weights_date": null, "openness": "open source", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fine-tuning recipe for OpenVLA reaching 97.1% average on LIBERO.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2502.19645" ], "unverified": [ "developer" ] }, { "id": "stanford-university-s1", "name": "s1", "level": "model", "family_id": null, "developer": [ "stanford-university", "university-of-washington", "allen-ai", "contextual-ai" ], "model_class": "reasoning", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Qwen2.5-32B-Instruct fine-tuned on 1,000 curated examples, 16 H100 for 26 minutes.", "context_length": null, "release_date": "2025-01", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Small-data reasoning fine-tune of Qwen2.5-32B-Instruct that raised AIME24 from 26.7 to 56.7.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2501.19393v1" ], "unverified": [] }, { "id": "stanford-university-satmae", "name": "SatMAE", "level": "model", "family_id": null, "developer": [ "stanford-university" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": "MAE ViT with temporal embedding and grouped spectral positional encodings", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-07-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Stanford multispectral/temporal MAE; evaluated on fMoW RGB and fMoW Sentinel.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2207.08051", "https://github.com/sustainlab-group/SatMAE" ], "unverified": [] }, { "id": "stepaudio-2-5-tts", "name": "StepAudio 2.5 TTS", "level": "model", "family_id": null, "developer": [], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-08", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "TTS model on the Artificial Analysis TTS arena; developer not shown.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/text-to-speech/leaderboard" ], "unverified": [] }, { "id": "stepfun-ace-step-1-5", "name": "ACE-Step 1.5", "level": "model", "family_id": null, "developer": [ "stepfun", "ace-studio" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "hybrid", "architecture_detail": "LM planner (Qwen3 0.6B / 1.7B / 4B) plus DiT", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01-31", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": "Under 4 GB VRAM", "runtimes": [], "languages_note": "50+ languages; Indonesian not confirmed", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open MIT music model jointly developed by StepFun and ACE Studio (community source), with a v1.5 XL Turbo version.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/ACE-Step/Ace-Step1.5", "https://huggingface.co/models?pipeline_tag=text-to-audio&sort=trending", "https://comfyui-wiki.com/en/models/ace-step" ], "unverified": [] }, { "id": "stepfun-ace-step-v1", "name": "ACE-Step v1", "level": "model", "family_id": null, "developer": [ "stepfun", "ace-studio" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First ACE-Step model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/ACE-Step/Ace-Step1.5" ], "unverified": [] }, { "id": "stepfun-step-3-5-flash", "name": "Step-3.5-Flash", "level": "model", "family_id": null, "developer": [ "stepfun" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-03", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/stepfun-ai/Step-3.7-Flash" ], "unverified": [] }, { "id": "stepfun-step-3-7-flash", "name": "Step-3.7-Flash", "level": "model", "family_id": null, "developer": [ "stepfun" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "Sparse MoE vision-language model (196B language backbone + 1.8B vision encoder)", "params_total": 198000000000, "params_active": 11000000000, "params_checkpoint": null, "params_note": "About 11B active", "context_length": 256000, "release_date": "2026-06-03", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "multi-GPU node", "single datacentre GPU" ], "footprint_note": "NVFP4 variant listed at 104B", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "StepFun's 198B sparse MoE vision-language model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/stepfun-ai/Step-3.7-Flash" ], "unverified": [] }, { "id": "stepfun-step-5-preview", "name": "Step 5 Preview", "level": "model", "family_id": null, "developer": [ "stepfun" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/leaderboards/models", "https://arena.ai/leaderboard/chat/text" ], "unverified": [ "release_date", "licence", "modalities_in" ] }, { "id": "stormer", "name": "Stormer (ENS)", "level": "model", "family_id": null, "developer": [ "ucla" ], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-12-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Ensemble weather model on WeatherBench 2 (2020).", "verified_on": "2026-10-10", "sources": [ "https://researchsites.withgoogle.com/weatherbench/deterministic/", "https://arxiv.org/html/2312.03876", "https://arxiv.org/abs/2312.03876" ], "unverified": [] }, { "id": "sub-pca", "name": "(Sub)-PCA", "level": "model", "family_id": null, "developer": [], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Simple statistical anomaly detection baseline, rank 2 on TSB-AD univariate.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2602.01359" ], "unverified": [] }, { "id": "suno", "name": "Suno", "level": "family", "family_id": null, "developer": [ "suno" ], "model_class": "music and audio", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Suno music generation versions.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Suno_(platform)" ], "unverified": [] }, { "id": "suno-bark", "name": "Bark", "level": "model", "family_id": null, "developer": [ "suno" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-04", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Suno's open TTS model, Apr 2023.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Suno_(platform)" ], "unverified": [] }, { "id": "suno-v3", "name": "Suno v3", "level": "model", "family_id": "suno", "developer": [ "suno" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-03", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Suno model version.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Suno_(platform)" ], "unverified": [] }, { "id": "suno-v4", "name": "Suno v4", "level": "model", "family_id": "suno", "developer": [ "suno" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-11", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Suno model version.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Suno_(platform)" ], "unverified": [] }, { "id": "suno-v4-5-all", "name": "Suno v4.5-all", "level": "model", "family_id": "suno", "developer": [ "suno" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Suno model version.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Suno_(platform)" ], "unverified": [] }, { "id": "suno-v5", "name": "Suno v5", "level": "model", "family_id": "suno", "developer": [ "suno" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Suno v5 (date unverified).", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Suno_(platform)" ], "unverified": [ "release_date" ] }, { "id": "suno-v5-5", "name": "Suno v5.5", "level": "model", "family_id": "suno", "developer": [ "suno" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-03-26", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Suno music model in the app and Suno Studio; superseded by Suno v6. Suno settled with Warner Music in Nov 2025 and trains on licensed WMG catalogue.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Suno_(platform)", "https://alphasignal.ai/news/artificial-analysis-music-arena-v1-1-crowns-suno-v6-across-1-000-harder-prompts" ], "unverified": [] }, { "id": "suno-v6", "name": "Suno v6", "level": "model", "family_id": "suno", "developer": [ "suno" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Suno model ranked first on both Artificial Analysis Music Arena v1.1 boards; release date unverified.", "verified_on": "2026-10-10", "sources": [ "https://alphasignal.ai/news/artificial-analysis-music-arena-v1-1-crowns-suno-v6-across-1-000-harder-prompts", "https://digg.com/ai/g8egglmo" ], "unverified": [ "release_date" ] }, { "id": "suno-v6-mini", "name": "Suno v6-mini", "level": "variant", "family_id": "suno-v6", "developer": [ "suno" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Smaller Suno v6 version.", "verified_on": "2026-10-10", "sources": [ "https://alphasignal.ai/news/artificial-analysis-music-arena-v1-1-crowns-suno-v6-across-1-000-harder-prompts", "https://digg.com/ai/g8egglmo" ], "unverified": [] }, { "id": "supertone-supertonic-3", "name": "Supertone supertonic-3", "level": "model", "family_id": null, "developer": [], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open TTS model trending on Hugging Face (2025 to 2026).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/models?pipeline_tag=text-to-speech&sort=trending" ], "unverified": [] }, { "id": "swe-bench-team-swe-agent-lm-32b", "name": "SWE-agent-LM-32B", "level": "model", "family_id": null, "developer": [ "swe-bench-team" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 33000000000, "params_active": null, "params_checkpoint": null, "params_note": "About 33B", "context_length": null, "release_date": "2025-04-30", "release_date_basis": "notes", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Research agent model fine-tuned on 5k SWE-smith trajectories.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/SWE-bench/SWE-agent-LM-32B" ], "unverified": [] }, { "id": "swiss-ai-apertus", "name": "Apertus", "level": "family", "family_id": null, "developer": [ "swiss-ai" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Apertus is the Swiss AI Initiative family of fully multilingual open language models.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/swiss-ai/Apertus-70B-Instruct-2509" ], "unverified": [] }, { "id": "swiss-ai-apertus-1-0", "name": "Apertus 1.0", "level": "model", "family_id": null, "developer": [ "swiss-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/swiss-ai/Apertus-v1.5-70B" ], "unverified": [] }, { "id": "swiss-ai-apertus-70b-2509", "name": "Apertus-70B-2509", "level": "model", "family_id": "swiss-ai-apertus", "developer": [ "swiss-ai" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Base pretrained Apertus 70B model named as the base of Apertus-70B-Instruct-2509.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/swiss-ai/Apertus-70B-Instruct-2509" ], "unverified": [] }, { "id": "swiss-ai-apertus-70b-instruct-2509", "name": "Apertus 70B Instruct (Apertus-70B-Instruct-2509)", "level": "model", "family_id": "swiss-ai-apertus", "developer": [ "swiss-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "Decoder-only transformer with xIELU activation, trained with the AdEMAMix optimizer", "params_total": 70000000000, "params_active": null, "params_checkpoint": null, "params_note": "Card says 70B; safetensors metadata lists 71B", "context_length": 65536, "release_date": "2025-09-02", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Card states 1811 natively supported languages; Indonesian not named", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Instruction-tuned 70B Apertus model from the Swiss AI Initiative, built for massively multilingual text generation with open weights.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/swiss-ai/Apertus-70B-Instruct-2509", "https://artificialanalysis.ai/models/apertus-70b-instruct" ], "unverified": [] }, { "id": "swiss-ai-apertus-8b-2509", "name": "Apertus-8B-2509", "level": "model", "family_id": "swiss-ai-apertus", "developer": [ "swiss-ai" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Base pretrained Apertus 8B model named as the base of Apertus-8B-Instruct-2509.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/swiss-ai/Apertus-8B-Instruct-2509" ], "unverified": [] }, { "id": "swiss-ai-apertus-8b-instruct-2509", "name": "Apertus 8B Instruct (Apertus-8B-Instruct-2509)", "level": "model", "family_id": "swiss-ai-apertus", "developer": [ "swiss-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "Decoder-only transformer with xIELU activation, trained from scratch on 15T tokens with the AdEMAMix optimizer", "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 65536, "release_date": "2025-09-02", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Card states 1811 natively supported languages; Indonesian not named", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Instruction-tuned 8B Apertus model from the Swiss AI Initiative for massively multilingual text generation with open weights.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/swiss-ai/Apertus-8B-Instruct-2509", "https://artificialanalysis.ai/models/apertus-8b-instruct" ], "unverified": [] }, { "id": "swiss-ai-apertus-v1-5", "name": "Apertus v1.5", "level": "family", "family_id": null, "developer": [ "swiss-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "fully open", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Swiss AI's fully open models (open weights, open data, full training recipes), Apache 2.0 plus an Acceptable Use Policy.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/swiss-ai", "https://huggingface.co/swiss-ai/Apertus-v1.5-70B" ], "unverified": [] }, { "id": "swiss-ai-apertus-v1-5-70b", "name": "Apertus v1.5 70B", "level": "model", "family_id": "swiss-ai-apertus-v1-5", "developer": [ "swiss-ai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "audio" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "Dense, xIELU activation, AdEMAMix optimiser", "params_total": 70000000000, "params_active": null, "params_checkpoint": null, "params_note": "72B listed", "context_length": 262144, "release_date": "2026-09", "release_date_basis": "registry date", "weights_date": null, "openness": "fully open", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "multi-GPU node", "single datacentre GPU" ], "footprint_note": "Single GPU at FP8", "runtimes": [], "languages_note": null, "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Swiss AI's fully open 70B model (continued pretraining of Apertus 1.0), Apache 2.0 plus an AUP. Finetunes 1, quants 7.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/swiss-ai", "https://huggingface.co/swiss-ai/Apertus-v1.5-70B", "https://leaderboard.sea-lion.ai/detailed/SEA", "https://leaderboard.sea-lion.ai/detailed/ID" ], "unverified": [ "indonesia_hosting" ], "indonesian_measured": [ { "benchmark": "SEA-HELM Indonesian leaderboard (ID aggregate, first score column of the Indonesian Competency and Task Scores table)", "score": "64.69 (95% CI -1.37 to +1.36)", "metric": "score 0 to 100, average of 2000 bootstraps with 95% CI", "run_by": "AI Singapore (SEA-HELM leaderboard maintainer)", "url": "https://leaderboard.sea-lion.ai/detailed/ID" } ] }, { "id": "swiss-ai-apertus-v1-5-8b", "name": "Apertus v1.5 8B", "level": "model", "family_id": "swiss-ai-apertus-v1-5", "developer": [ "swiss-ai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "audio" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 262144, "release_date": "2026-07-24", "release_date_basis": "registry date", "weights_date": null, "openness": "fully open", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Swiss AI's fully open 8B model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/swiss-ai" ], "unverified": [] }, { "id": "swivid-f5-tts", "name": "F5-TTS", "level": "model", "family_id": null, "developer": [ "swivid" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "flow matching", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-10-09", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open flow-matching TTS model (v1 Base), non-commercial.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/SWivid/F5-TTS" ], "unverified": [] }, { "id": "synapxe-ace-ai", "name": "ACE-AI", "level": "model", "family_id": null, "developer": [ "synapxe" ], "model_class": "tabular", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Cardiovascular risk model listed by Synapxe (Singapore).", "verified_on": "2026-10-10", "sources": [ "https://www.synapxe.sg/healthtech/health-ai" ], "unverified": [ "indonesian_measured", "indonesia_hosting", "model_class" ] }, { "id": "synapxe-asar", "name": "ASAR", "level": "model", "family_id": null, "developer": [ "synapxe" ], "model_class": "tabular", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Drug safety signal detection model listed by Synapxe (Singapore); no model details given.", "verified_on": "2026-10-10", "sources": [ "https://www.synapxe.sg/healthtech/health-ai" ], "unverified": [ "model_class" ] }, { "id": "synapxe-cape", "name": "CAPE", "level": "model", "family_id": null, "developer": [ "synapxe" ], "model_class": "tabular", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Pneumonia severity model listed by Synapxe (Singapore).", "verified_on": "2026-10-10", "sources": [ "https://www.synapxe.sg/healthtech/health-ai" ], "unverified": [ "indonesian_measured", "indonesia_hosting", "model_class" ] }, { "id": "synapxe-selena-plus", "name": "SELENA+", "level": "model", "family_id": null, "developer": [ "synapxe" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Deep-learning eye screening model listed by Synapxe, Singapore's national health IT agency.", "verified_on": "2026-10-10", "sources": [ "https://www.synapxe.sg/healthtech/health-ai" ], "unverified": [] }, { "id": "synthesia-3-0", "name": "Synthesia 3.0", "level": "model", "family_id": null, "developer": [ "synthesia" ], "model_class": "avatar", "modalities_in": [ "text" ], "modalities_out": [ "video", "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "140 languages and accents; dubbing in 30 languages with lip-sync", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Synthesia avatar video SaaS; interactive avatar agents planned for 2026.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Synthesia_(company)" ], "unverified": [] }, { "id": "systran-faster-whisper-small", "name": "faster-whisper-small", "level": "variant", "family_id": "openai-whisper", "developer": [ "systran" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "autoregressive", "architecture_detail": "openai/whisper-small converted to the CTranslate2 format with FP16 weights", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-11-23", "release_date_basis": "registry date", "weights_date": "2023-11-23", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Card language tags include Indonesian, Javanese (jw), Sundanese, Malay, Tagalog, Thai and Vietnamese", "indonesian_support": "listed by developer", "regional_languages": [ "Javanese", "Sundanese" ], "indonesia_hosting": "unknown", "status": null, "summary": "CTranslate2 conversion of OpenAI whisper-small for use with CTranslate2 or the faster-whisper library.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/Systran/faster-whisper-small", "https://huggingface.co/api/models/Systran/faster-whisper-small" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "systran-faster-whisper-tiny", "name": "faster-whisper-tiny", "level": "variant", "family_id": "openai-whisper", "developer": [ "systran" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "autoregressive", "architecture_detail": "openai/whisper-tiny converted to the CTranslate2 format with FP16 weights", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-11-23", "release_date_basis": "registry date", "weights_date": "2023-11-23", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Card language tags include Indonesian, Javanese (jw), Sundanese, Malay, Tagalog, Thai and Vietnamese", "indonesian_support": "listed by developer", "regional_languages": [ "Javanese", "Sundanese" ], "indonesia_hosting": "unknown", "status": null, "summary": "CTranslate2 conversion of OpenAI whisper-tiny for use with CTranslate2 or the faster-whisper library.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/Systran/faster-whisper-tiny", "https://huggingface.co/api/models/Systran/faster-whisper-tiny" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "tabm", "name": "TabM", "level": "model", "family_id": null, "developer": [ "yandex" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-10-31", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Deep tabular model on TabArena.", "verified_on": "2026-10-10", "sources": [ "https://benchmarklist.com/benchmarks/tabarena_all/", "https://arxiv.org/html/2410.24210", "https://arxiv.org/abs/2410.24210" ], "unverified": [] }, { "id": "tace-v1-oam-m", "name": "TACE-v1-OAM-M", "level": "model", "family_id": null, "developer": [], "model_class": "materials", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "GNN", "architecture_detail": "Equivariant GNN interatomic potential (MLIP)", "params_total": 18800000, "params_active": null, "params_checkpoint": null, "params_note": "Training data OMat24+sAlex+MPtrj", "context_length": null, "release_date": "2026-01-06", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Machine-learned interatomic potential on Matbench Discovery (rank 8 by CPS, date added 2026-01-06).", "verified_on": "2026-10-10", "sources": [ "https://matbench-discovery.materialsproject.org/" ], "unverified": [ "developer" ] }, { "id": "tencent-auk", "name": "AuK", "level": "model", "family_id": null, "developer": [ "tencent" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open TTS model trending on Hugging Face (2025 to 2026).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/models?pipeline_tag=text-to-speech&sort=trending" ], "unverified": [] }, { "id": "tencent-hunyuan3d-2-1", "name": "Hunyuan3D-2.1", "level": "model", "family_id": null, "developer": [ "tencent" ], "model_class": "3D", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-06-18", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "tencent-hunyuan-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tencent image-to-3D model with production PBR materials (also 2, 2mv, Omni).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/tencent/Hunyuan3D-2.1", "https://huggingface.co/models?pipeline_tag=image-to-3d&sort=trending" ], "unverified": [] }, { "id": "tencent-hunyuanimage", "name": "HunyuanImage", "level": "family", "family_id": null, "developer": [ "tencent" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tencent Hunyuan image model line.", "verified_on": "2026-10-10", "sources": [ "https://comfyui-wiki.com/news/2025-09-27-tencent-open-source-hunyuan-image-3-0" ], "unverified": [] }, { "id": "tencent-hunyuanimage-3-0", "name": "HunyuanImage 3.0", "level": "model", "family_id": "tencent-hunyuanimage", "developer": [ "tencent" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "MoE", "architecture_detail": "Autoregressive MoE plus Transfusion, 64 experts", "params_total": 80000000000, "params_active": 13000000000, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-09-28", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights restricted", "licence": "tencent-hunyuan-community", "access_routes": [ "open weights download" ], "deployment_tiers": [ "multi-GPU node" ], "footprint_note": "3 to 4 x 80 GB GPUs", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tencent open-weight autoregressive MoE image model under the Hunyuan Community licence, which excludes the EU, UK and South Korea.", "verified_on": "2026-10-10", "sources": [ "https://comfyui-wiki.com/news/2025-09-27-tencent-open-source-hunyuan-image-3-0", "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text", "https://www.tencent.com/en-us/about.html" ], "unverified": [] }, { "id": "tencent-hunyuanimage-3-0-instruct", "name": "HunyuanImage 3.0 Instruct", "level": "model", "family_id": "tencent-hunyuanimage", "developer": [ "tencent" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-09-25", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Instruct version of HunyuanImage 3.0 on Artificial Analysis.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text", "https://comfyui-wiki.com/news/2025-09-27-tencent-open-source-hunyuan-image-3-0", "https://huggingface.co/tencent/HunyuanImage-3.0-Instruct" ], "unverified": [] }, { "id": "tencent-hunyuanimage-3-5-preview", "name": "HunyuanImage 3.5 Preview", "level": "model", "family_id": "tencent-hunyuanimage", "developer": [ "tencent" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": "Preview of HunyuanImage 3.5 on Artificial Analysis.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text", "https://comfyui-wiki.com/news/2025-09-27-tencent-open-source-hunyuan-image-3-0" ], "unverified": [] }, { "id": "tencent-hunyuanvideo", "name": "HunyuanVideo", "level": "model", "family_id": "tencent-hunyuanvideo-family", "developer": [ "tencent" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": 13000000000, "params_active": null, "params_checkpoint": null, "params_note": "13B, unverified", "context_length": null, "release_date": "2024", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First HunyuanVideo model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/tencent/HunyuanVideo-1.5" ], "unverified": [ "params_total" ] }, { "id": "tencent-hunyuanvideo-1-5", "name": "HunyuanVideo 1.5", "level": "model", "family_id": "tencent-hunyuanvideo-family", "developer": [ "tencent" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": "DiT plus 3D causal VAE, SSTA attention", "params_total": 8300000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-11-20", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "tencent-hunyuan-community", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": "14 GB with offload; RTX 4090 renders 480p in 75 s", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tencent 8.3B open-weight video model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/tencent/HunyuanVideo-1.5", "https://huggingface.co/models?pipeline_tag=text-to-video&sort=trending" ], "unverified": [] }, { "id": "tencent-hunyuanvideo-family", "name": "HunyuanVideo", "level": "family", "family_id": null, "developer": [ "tencent" ], "model_class": "video generation", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tencent Hunyuan video line.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/tencent/HunyuanVideo-1.5" ], "unverified": [] }, { "id": "tencent-hunyuanvideo-foley", "name": "HunyuanVideo-Foley", "level": "model", "family_id": null, "developer": [ "tencent" ], "model_class": "music and audio", "modalities_in": [ "video" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-09", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tencent video-to-sound-effects model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/models?pipeline_tag=text-to-audio&sort=trending" ], "unverified": [] }, { "id": "tencent-hunyuanworld-1-0", "name": "HunyuanWorld 1.0", "level": "model", "family_id": "tencent-hy-world", "developer": [ "tencent" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-07", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First HunyuanWorld model, Jul 2025.", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/hy_world_2" ], "unverified": [] }, { "id": "tencent-hunyuanworld-voyager", "name": "HunyuanWorld-Voyager", "level": "model", "family_id": "tencent-hy-world", "developer": [ "tencent" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-09", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tencent world model, Sep 2025.", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/hy_world_2" ], "unverified": [] }, { "id": "tencent-hy-world", "name": "HY-World", "level": "family", "family_id": null, "developer": [ "tencent" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tencent Hunyuan world model line.", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/hy_world_2" ], "unverified": [] }, { "id": "tencent-hy-world-1-5", "name": "HY-World 1.5", "level": "model", "family_id": "tencent-hy-world", "developer": [ "tencent" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Video-based HY-World version.", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/hy_world_2" ], "unverified": [] }, { "id": "tencent-hy-world-2-0", "name": "HY-World 2.0", "level": "model", "family_id": "tencent-hy-world", "developer": [ "tencent" ], "model_class": "world model", "modalities_in": [ "text", "image" ], "modalities_out": [ "3D" ], "architecture": "hybrid", "architecture_detail": "Pipeline: HY-Pano 2.0 (~80B, text/image to 360 panorama), WorldNav (trajectory planning), WorldStereo 2.0 (~17B), WorldMirror 2.0 (~1.2B); outputs mesh, 3DGS, point cloud", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Components: HY-Pano 2.0 ~80B, WorldStereo 2.0 ~17B, WorldMirror 2.0 ~1.2B", "context_length": null, "release_date": "2026-04-16", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights restricted", "licence": "tencent-hy-world-2-0-community", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": "Needs A100/H100 40GB+", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tencent open-weight 3D world generation pipeline; HY-Pano 2.0 weights 11 May and WorldStereo 2.0 weights 18 May 2026. Claimed #1 open-source on WorldScore without a number.", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/hy_world_2", "https://blog.buildfastwithai.com/tencent-alibaba-world-models-april-2026" ], "unverified": [] }, { "id": "tencent-hy3", "name": "Hy3", "level": "model", "family_id": null, "developer": [ "tencent" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "MoE (192 experts, top-8 active) plus a 3.8B MTP layer", "params_total": 295000000000, "params_active": 21000000000, "params_checkpoint": 298786155776, "params_note": "Plus a 3.8B MTP layer, which brings the checkpoint to about 299B", "context_length": 256000, "release_date": "2026-07-06", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "API", "open weights download" ], "deployment_tiers": [ "multi-GPU node" ], "footprint_note": "FP8 variant", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tencent Hunyuan's text-only open-weight MoE model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/tencent/Hy3/raw/main/README.md", "https://huggingface.co/tencent/Hy3/raw/main/LICENSE", "https://huggingface.co/api/models/tencent/Hy3/commits/main", "https://www.news.cn/tech/20260706/d611426ecba54475b2dd94920e4e3557/c.html", "https://artificialanalysis.ai/models/hy3", "https://arena.ai/leaderboard/chat/text", "https://huggingface.co/tencent/Hy3" ], "unverified": [] }, { "id": "tencent-hy4-preview", "name": "Hy4-preview", "level": "model", "family_id": null, "developer": [ "tencent" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "MoE (256 routed experts, 8 active + 1 shared), Gated DSA attention, iHC", "params_total": 770000000000, "params_active": 49000000000, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": "2026-08-28", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "multi-GPU node" ], "footprint_note": "FP8 variant", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": "Tencent Hunyuan's 770B MoE preview text model. The card names no territorial exclusions; full licence text not reviewed.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/tencent/Hy4-preview" ], "unverified": [] }, { "id": "tesla-fsd-v14", "name": "Tesla FSD v14", "level": "model", "family_id": null, "developer": [ "tesla" ], "model_class": "driving", "modalities_in": [ "video" ], "modalities_out": [ "action" ], "architecture": null, "architecture_detail": "End-to-end pixels to controls, with intermediate tokens and generative Gaussian splatting for interpretability", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tesla end-to-end driving model (v14.3.2 in update 2026.2.9.8), trained and evaluated with the Tesla World Simulator.", "verified_on": "2026-10-10", "sources": [ "https://www.itiger.com/news/1146546403", "https://www.notateslaapp.com/de/software-updates/version/2026.2.9.8/versionshinweise" ], "unverified": [] }, { "id": "tesla-optimus-v3", "name": "Tesla Optimus V3", "level": "model", "family_id": null, "developer": [ "tesla" ], "model_class": "robot policy", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "internal only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "announced", "summary": "Tesla humanoid (V3 / Gen 3), reveal pending as of Sep 2026; shares the FSD World Simulator and uses Grok for voice. Musk said in Jan 2026 it was still in R&D.", "verified_on": "2026-10-10", "sources": [ "https://www.itiger.com/news/1146546403", "https://optimusk.blog/blog/tesla-optimus-humanoid-robot-latest-version-2026/" ], "unverified": [ "status" ] }, { "id": "tesla-world-simulator", "name": "Tesla World Simulator", "level": "model", "family_id": null, "developer": [ "tesla" ], "model_class": "world model", "modalities_in": [ "action", "sensor" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": "Neural digital twin generating multi-camera driving video from vehicle state and commands", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10-26", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "internal only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tesla simulator used for closed-loop FSD evaluation, near-miss replay and adversarial testing, also applied to Optimus.", "verified_on": "2026-10-10", "sources": [ "https://www.itiger.com/news/1146546403" ], "unverified": [] }, { "id": "thinking-machines-inkling", "name": "Inkling", "level": "model", "family_id": null, "developer": [ "thinking-machines" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Thinking Machines model with open weights per steel.dev.", "verified_on": "2026-10-10", "sources": [ "https://leaderboard.steel.dev/leaderboards/swe-bench-verified.md" ], "unverified": [ "openness" ] }, { "id": "tianji", "name": "Tianji", "level": "model", "family_id": null, "developer": [], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Weather model on WeatherBench 2 (2022).", "verified_on": "2026-10-10", "sources": [ "https://researchsites.withgoogle.com/weatherbench/deterministic/" ], "unverified": [] }, { "id": "tii-falcon-e-3b", "name": "Falcon-E 3B", "level": "model", "family_id": null, "developer": [ "tii" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 3000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights restricted", "licence": "falcon-llm-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/tiiuae" ], "unverified": [] }, { "id": "tii-falcon-h1-7b-base", "name": "Falcon-H1-7B-Base", "level": "model", "family_id": null, "developer": [ "tii" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 7000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/tiiuae/Falcon-H1R-7B" ], "unverified": [] }, { "id": "tii-falcon-h1-arabic", "name": "Falcon-H1-Arabic", "level": "model", "family_id": null, "developer": [ "tii" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01-05", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "falcon-llm-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Arabic", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/tiiuae" ], "unverified": [] }, { "id": "tii-falcon-h1-tiny", "name": "Falcon-H1-Tiny", "level": "model", "family_id": null, "developer": [ "tii" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights restricted", "licence": "falcon-llm-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/tiiuae" ], "unverified": [] }, { "id": "tii-falcon-h1r-7b", "name": "Falcon-H1R-7B", "level": "model", "family_id": null, "developer": [ "tii" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Hybrid Transformer + Mamba2", "params_total": 7000000000, "params_active": null, "params_checkpoint": null, "params_note": "8B listed", "context_length": 262000, "release_date": "2026-01-05", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "falcon-llm-license", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU", "CPU server" ], "footprint_note": "262K is the vLLM default", "runtimes": [ "vllm" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "TII's 7B hybrid reasoning model; no Falcon-H2 found.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/tiiuae", "https://huggingface.co/tiiuae/Falcon-H1R-7B" ], "unverified": [] }, { "id": "tinyllama-1-1b", "name": "TinyLlama 1.1B", "level": "model", "family_id": null, "developer": [ "sutd" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "1.1B", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "CPU server", "laptop", "mobile or NPU" ], "footprint_note": "Ollama GGUF build", "runtimes": [ "ollama" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small model listed in the Ollama library for CPU, laptop and phone use.", "verified_on": "2026-10-10", "sources": [ "https://ollama.com/library", "https://arxiv.org/html/2401.02385" ], "unverified": [] }, { "id": "tomoviee-2-0", "name": "ToMoviee 2.0", "level": "model", "family_id": null, "developer": [], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Video model listed on the VBench leaderboard; developer not stated in the notes.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/spaces/Vchitect/VBench_Leaderboard" ], "unverified": [] }, { "id": "tsinghua-university-dinomaly2", "name": "Dinomaly2", "level": "model", "family_id": null, "developer": [ "tsinghua-university", "beijing-institute-of-technology", "shanghai-jiao-tong-university", "fudan-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": "ViT-S/B/L on DINOv2-R features; multi-class unified unsupervised anomaly detection", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10-24", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "research only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Research model for industrial visual anomaly detection; image AUROC MVTec AD 99.9, VisA 99.3, Real-IAD 92.1. Licence unverified.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2510.17611" ], "unverified": [ "licence", "indonesian_measured", "indonesia_hosting" ] }, { "id": "tsinghua-university-sundial-base-128m", "name": "Sundial (Timer 3.0) base-128m", "level": "model", "family_id": null, "developer": [ "tsinghua-university" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "flow matching", "architecture_detail": "Decoder-only causal transformer, 16-step patches, TimeFlow (flow-matching) loss; output 720; 12 layers", "params_total": 128000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 2880, "release_date": "2025-02-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU", "CPU server" ], "footprint_note": "Integrated into Apache IoTDB Aug 2025", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tsinghua THUML generative time-series foundation model (ICML 2025 Oral). Context 2,880 time steps.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/thuml/sundial-base-128m", "https://huggingface.co/spaces/Salesforce/GIFT-Eval" ], "unverified": [] }, { "id": "tsinghua-university-timer-s1", "name": "Timer-S1", "level": "model", "family_id": null, "developer": [ "tsinghua-university", "bytedance" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": "Timer lineage", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tsinghua and ByteDance time-series model (per GIFT-Eval config); sibling of Sundial.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2605.20119", "https://huggingface.co/spaces/Salesforce/GIFT-Eval" ], "unverified": [ "release_date" ] }, { "id": "tsinghua-university-video-salmonn-2-plus-72b", "name": "video-SALMONN 2+ 72B", "level": "model", "family_id": null, "developer": [ "tsinghua-university", "bytedance" ], "model_class": "multimodal LLM", "modalities_in": [ "video", "audio", "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "72B", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tsinghua and ByteDance video-understanding model top on Video-MME with subtitles.", "verified_on": "2026-10-10", "sources": [ "https://video-mme.github.io/home_page.html" ], "unverified": [] }, { "id": "tsinghua-university-yolov10", "name": "YOLOv10", "level": "model", "family_id": null, "developer": [ "tsinghua-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image", "text" ], "architecture": "CNN", "architecture_detail": "NMS-free YOLO", "params_total": 2800000, "params_active": null, "params_checkpoint": null, "params_note": "2.80M parameters for the n variant (HF jameslahm/yolov10n)", "context_length": null, "release_date": "2024-05", "release_date_basis": "announcement page", "weights_date": "2024-06-01", "openness": "open source", "licence": "agpl-3-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "embedded GPU", "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tsinghua University (School of Software and Department of Automation) NMS-free YOLO detector; source of the NMS-free design used in YOLO26.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2405.14458", "https://huggingface.co/api/models/jameslahm/yolov10n", "https://docs.ultralytics.com/yolo/" ], "unverified": [] }, { "id": "ugm-tbscreen-ai", "name": "TBScreen.AI", "level": "model", "family_id": null, "developer": [ "universitas-gadjah-mada", "university-of-melbourne" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "research only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": "Chest X-ray TB screening CAD built on Indonesian data by UGM and the University of Melbourne (KONEKSI-funded); reported sensitivity 83.16%. Online version operable, offline version planned.", "verified_on": "2026-10-10", "sources": [ "https://www.antaranews.com/berita/5678115/kemenkes-teknologi-tbscreenai-mampu-tingkatkan-efisiensi-skrining-tb" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "ultralytics-yolo11", "name": "YOLO11", "level": "model", "family_id": null, "developer": [ "ultralytics" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image", "text" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-10-16", "release_date_basis": "registry date", "weights_date": null, "openness": "open source", "licence": "agpl-3-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "embedded GPU", "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier Ultralytics multi-task YOLO model, described as the mature option for stable production.", "verified_on": "2026-10-10", "sources": [ "https://docs.ultralytics.com/yolo/", "https://arxiv.org/html/2511.09554v2", "https://huggingface.co/api/models/Ultralytics/YOLO11", "https://docs.ultralytics.com/models/yolo26/", "https://journal.ugm.ac.id/ijccs/article/view/111253" ], "unverified": [] }, { "id": "ultralytics-yolo26", "name": "YOLO26", "level": "family", "family_id": null, "developer": [ "ultralytics" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image", "text" ], "architecture": "CNN", "architecture_detail": "CNN YOLO; DFL removed; optional NMS-free end-to-end head; MuSGD optimiser (SGD + Muon hybrid); ProgLoss + STAL", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "2.4M (n) to 55.7M (x) fused", "context_length": null, "release_date": "2026-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "agpl-3-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server", "laptop", "embedded GPU", "single datacentre GPU" ], "footprint_note": "n: 38.9 ms CPU ONNX; x: 11.8 ms T4 TensorRT. Also available under an Ultralytics Enterprise licence. Tier fit is a judgement.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Ultralytics real-time vision model family (n/s/m/l/x) for detection, segmentation, semantic segmentation, depth, classification, pose, OBB and open-vocabulary variants. Follows YOLO11; the NMS-free idea is traced to YOLOv10.", "verified_on": "2026-10-10", "sources": [ "https://docs.ultralytics.com/yolo/", "https://docs.ultralytics.com/models/yolo26/", "https://ultralytics.com/about" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "ultralytics-yolo26-l", "name": "YOLO26l", "level": "variant", "family_id": "ultralytics-yolo26", "developer": [ "ultralytics" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image", "text" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "agpl-3-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU", "embedded GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "YOLO26 size variant l.", "verified_on": "2026-10-10", "sources": [ "https://docs.ultralytics.com/models/yolo26/" ], "unverified": [] }, { "id": "ultralytics-yolo26-m", "name": "YOLO26m", "level": "variant", "family_id": "ultralytics-yolo26", "developer": [ "ultralytics" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image", "text" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "agpl-3-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU", "embedded GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "YOLO26 size variant m.", "verified_on": "2026-10-10", "sources": [ "https://docs.ultralytics.com/models/yolo26/" ], "unverified": [] }, { "id": "ultralytics-yolo26-n", "name": "YOLO26n", "level": "variant", "family_id": "ultralytics-yolo26", "developer": [ "ultralytics" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image", "text" ], "architecture": "CNN", "architecture_detail": null, "params_total": 2400000, "params_active": null, "params_checkpoint": null, "params_note": "fused", "context_length": null, "release_date": "2026-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "agpl-3-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server", "laptop", "embedded GPU" ], "footprint_note": "38.9 ms CPU ONNX", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "YOLO26 size variant n.", "verified_on": "2026-10-10", "sources": [ "https://docs.ultralytics.com/models/yolo26/" ], "unverified": [] }, { "id": "ultralytics-yolo26-s", "name": "YOLO26s", "level": "variant", "family_id": "ultralytics-yolo26", "developer": [ "ultralytics" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image", "text" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "agpl-3-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU", "embedded GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "YOLO26 size variant s.", "verified_on": "2026-10-10", "sources": [ "https://docs.ultralytics.com/models/yolo26/" ], "unverified": [] }, { "id": "ultralytics-yolo26-x", "name": "YOLO26x", "level": "variant", "family_id": "ultralytics-yolo26", "developer": [ "ultralytics" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image", "text" ], "architecture": "CNN", "architecture_detail": null, "params_total": 55700000, "params_active": null, "params_checkpoint": null, "params_note": "fused", "context_length": null, "release_date": "2026-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "agpl-3-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": "11.8 ms T4 TensorRT", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "YOLO26 size variant x.", "verified_on": "2026-10-10", "sources": [ "https://docs.ultralytics.com/models/yolo26/" ], "unverified": [] }, { "id": "ultralytics-yolo27", "name": "YOLO27", "level": "model", "family_id": null, "developer": [ "ultralytics" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": null, "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "announced", "summary": "Successor to YOLO26 described by Ultralytics as in final R&D, with no date given.", "verified_on": "2026-10-10", "sources": [ "https://docs.ultralytics.com/yolo/" ], "unverified": [] }, { "id": "ultralytics-yolov11-large", "name": "YOLOv11 Large", "level": "model", "family_id": null, "developer": [ "ultralytics" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Ultralytics object-detection model used in the MLPerf Inference v6.0 edge object-detection test.", "verified_on": "2026-10-10", "sources": [ "https://www.globenewswire.com/news-release/2026/04/01/3266801/0/en/mlcommons-releases-new-mlperf-inference-v6-0-benchmark-results.html", "https://github.com/ultralytics/ultralytics" ], "unverified": [] }, { "id": "ultralytics-yolov5", "name": "YOLOv5", "level": "model", "family_id": null, "developer": [ "ultralytics" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Ultralytics model listed on PyTorch Hub.", "verified_on": "2026-10-10", "sources": [ "https://pytorch.org/hub/", "https://github.com/ultralytics/yolov5" ], "unverified": [] }, { "id": "ultralytics-yolov8", "name": "YOLOv8", "level": "model", "family_id": null, "developer": [ "ultralytics" ], "model_class": "vision", "modalities_in": [], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Ultralytics detector used in a Gunadarma University FFB grading web app.", "verified_on": "2026-10-10", "sources": [ "https://ejournal.gunadarma.ac.id/index.php/tekno/article/view/67" ], "unverified": [] }, { "id": "uncharted-labs-udio", "name": "Udio", "level": "family", "family_id": null, "developer": [ "uncharted-labs" ], "model_class": "music and audio", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Udio music generation (beta Apr 2024, v1.5 Jul 2024, Allegro Mar 2025).", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Udio" ], "unverified": [] }, { "id": "uncharted-labs-udio-allegro", "name": "Udio Allegro", "level": "model", "family_id": "uncharted-labs-udio", "developer": [ "uncharted-labs" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Udio music model; after a UMG settlement (Oct 2025) the platform moves to training only on licensed music.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Udio", "https://www.builtinla.com/company/udio-udiocom" ], "unverified": [] }, { "id": "uncharted-labs-udio-v1-5", "name": "Udio v1.5", "level": "model", "family_id": "uncharted-labs-udio", "developer": [ "uncharted-labs" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-07", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Udio model version.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Udio" ], "unverified": [] }, { "id": "united-imaging-pulmonary-nodule-software", "name": "United Imaging pulmonary nodule software", "level": "model", "family_id": null, "developer": [ "united-imaging" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": "CNN detection software as a medical device for lung CT", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Lung CT nodule detection software. United Imaging says over 2,000 Chinese hospitals use it (manufacturer claim).", "verified_on": "2026-10-10", "sources": [ "https://pmc.ncbi.nlm.nih.gov/articles/PMC12998604" ], "unverified": [] }, { "id": "unitree-unifolm-vla-0", "name": "UnifoLM-VLA-0", "level": "model", "family_id": null, "developer": [ "unitree" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": null, "architecture_detail": "VLA initialised from UnifoLM-VLM-Base; code inherited from Qwen2.5-VL", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01-29", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Unitree open VLA with a released LIBERO checkpoint (no score) and one policy across 12 real task categories; licence not stated.", "verified_on": "2026-10-10", "sources": [ "https://raw.githubusercontent.com/unitreerobotics/unifolm-vla/main/README.md" ], "unverified": [] }, { "id": "unitree-unifolm-vlm-base", "name": "UnifoLM-VLM-Base", "level": "model", "family_id": null, "developer": [ "unitree" ], "model_class": "multimodal LLM", "modalities_in": [ "image", "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Unitree VLM fine-tuned on VQA and open robot data, base of UnifoLM-VLA-0.", "verified_on": "2026-10-10", "sources": [ "https://raw.githubusercontent.com/unitreerobotics/unifolm-vla/main/README.md" ], "unverified": [] }, { "id": "unitree-unifolm-wma-0", "name": "UnifoLM-WMA-0", "level": "model", "family_id": null, "developer": [ "unitree" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Unitree world-model-action repo (no details fetched).", "verified_on": "2026-10-10", "sources": [ "https://raw.githubusercontent.com/unitreerobotics/unifolm-vla/main/README.md" ], "unverified": [] }, { "id": "university-at-buffalo-yolo12", "name": "YOLO12", "level": "model", "family_id": null, "developer": [ "university-at-buffalo", "university-of-chinese-academy-of-sciences" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image", "text" ], "architecture": "other", "architecture_detail": "Attention-centric real-time object detector", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-02-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "agpl-3-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Attention-centric real-time object detector (detection only, pretrained) listed in Ultralytics docs.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2502.12524", "https://github.com/sunsmarterjie/yolov12", "https://docs.ultralytics.com/yolo/" ], "unverified": [] }, { "id": "upn-veteran-jawa-timur-yolov11-ffb-grader", "name": "YOLOv11 FFB ripeness grader (IJCCS 2025)", "level": "model", "family_id": null, "developer": [ "upn-veteran-jawa-timur" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10-31", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "research only" ], "deployment_tiers": [ "CPU server", "offline capable" ], "footprint_note": "ONNX Runtime on CPU; 255 to 300 ms latency", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Academic oil palm fresh fruit bunch ripeness detector on YOLOv11; 90.2% mAP@50. No plantation deployment stated. Affiliation unverified.", "verified_on": "2026-10-10", "sources": [ "https://journal.ugm.ac.id/ijccs/article/view/111253" ], "unverified": [ "developer", "indonesian_measured", "indonesia_hosting" ] }, { "id": "upstage-solar-mini-4", "name": "Solar Mini 4", "level": "model", "family_id": null, "developer": [ "upstage" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": 3000000000, "params_checkpoint": null, "params_note": "About 3B active", "context_length": null, "release_date": "2026-10-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://www.upstage.ai/blog" ], "unverified": [] }, { "id": "upstage-solar-open-100b", "name": "Solar-Open-100B", "level": "model", "family_id": null, "developer": [ "upstage" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 100000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/upstage/Solar-Open2-250B", "https://huggingface.co/upstage" ], "unverified": [] }, { "id": "upstage-solar-open-2", "name": "Solar Open 2", "level": "model", "family_id": null, "developer": [ "upstage" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "Hybrid-attention MoE (softmax + linear attention, 320 routed + 1 shared experts), NoPE", "params_total": 250000000000, "params_active": 15000000000, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": "2026-07-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "upstage-solar-license", "access_routes": [ "open weights download" ], "deployment_tiers": [ "multi-GPU node" ], "footprint_note": null, "runtimes": [], "languages_note": "English, Korean, Japanese", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Korea sovereign model (open).", "verified_on": "2026-10-10", "sources": [ "https://www.upstage.ai/blog", "https://huggingface.co/upstage/Solar-Open2-250B", "https://huggingface.co/upstage" ], "unverified": [] }, { "id": "upstage-solar-pro", "name": "Solar Pro", "level": "family", "family_id": null, "developer": [ "upstage" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Solar Pro is the Upstage line of API language models.", "verified_on": "2026-10-11", "sources": [ "https://console.upstage.ai/docs/models", "https://console.upstage.ai/docs/models/solar-pro-3" ], "unverified": [] }, { "id": "upstage-solar-pro-3", "name": "Solar Pro 3", "level": "model", "family_id": "upstage-solar-pro", "developer": [ "upstage" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 102000000000, "params_active": 12000000000, "params_checkpoint": null, "params_note": null, "context_length": 128000, "release_date": "2026-01-26", "release_date_basis": "inferred from page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "On-premises deployment by contacting Upstage", "runtimes": [], "languages_note": "Korean, English, Japanese", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Upstage 102B MoE API model with reasoning mode and tool calling; Upstage schedules full deprecation on 30 October 2026 in favour of Solar Pro 4.", "verified_on": "2026-10-11", "sources": [ "https://console.upstage.ai/docs/models/solar-pro-3", "https://console.upstage.ai/docs/models", "https://openrouter.ai/api/v1/models" ], "unverified": [] }, { "id": "upstage-solar-pro-4", "name": "Solar Pro 4", "level": "model", "family_id": null, "developer": [ "upstage" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-08-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": "API via Upstage Console, AWS, Azure, Snowflake, on-prem. AA lists a 'Solar Pro 45' entry with 512k context.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Upstage agentic model with tool calling and terminal tasks.", "verified_on": "2026-10-10", "sources": [ "https://www.upstage.ai/blog", "https://arena.ai/leaderboard/chat/text", "https://artificialanalysis.ai/leaderboards/models", "https://www.benchlm.ai/benchmarks/livecodebench" ], "unverified": [] }, { "id": "useful-sensors-moonshine", "name": "Moonshine (Moonshine Voice)", "level": "model", "family_id": null, "developer": [ "useful-sensors" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "Encoder-decoder for edge", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "27M to 331M", "context_length": null, "release_date": "2024", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server", "embedded GPU", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": "English-focused", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Edge ASR model; LICENSE contains MIT text and also mentions non-commercial terms, so per-model terms are unverified.", "verified_on": "2026-10-10", "sources": [ "https://www.gladia.io/blog/best-open-source-speech-to-text-models", "https://github.com/usefulsensors/moonshine", "https://github.com/usefulsensors/moonshine" ], "unverified": [ "licence", "release_date" ] }, { "id": "ustc-d-fine", "name": "D-FINE", "level": "model", "family_id": null, "developer": [ "university-of-science-and-technology-of-china", "hefei-institute-of-artificial-intelligence" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image", "text" ], "architecture": "dense transformer", "architecture_detail": "Real-time DETR", "params_total": 62900000, "params_active": null, "params_checkpoint": null, "params_note": "62.9M for dfine-xlarge-coco", "context_length": null, "release_date": "2024-10-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time DETR detector from USTC; RT-DETR lineage is unverified.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/abs/2410.13842", "https://huggingface.co/api/models/ustc-community/dfine-xlarge-coco", "https://arxiv.org/html/2511.09554v2" ], "unverified": [] }, { "id": "utopai-studios-utopai-x", "name": "Utopai X", "level": "model", "family_id": null, "developer": [ "utopai-studios" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-09", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Utopai Studios video model, 2nd on Artificial Analysis text-to-video.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/video/leaderboard/text-to-video", "https://www.utopaistudios.com/about" ], "unverified": [] }, { "id": "vast-tripo-3-0", "name": "Tripo 3.0", "level": "model", "family_id": null, "developer": [ "vast" ], "model_class": "3D", "modalities_in": [ "text", "image" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-08", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tripo AI 3D generation service; current version unverified.", "verified_on": "2026-10-10", "sources": [ "https://www-web.itiger.com/news/1138581647", "https://investgame.net/news/matrix-partners-china-has-led-a-446m-series-b-and-b-into-vast-maker-of-tripo-ai-in-its-fourth-round-of-2026/" ], "unverified": [ "status" ] }, { "id": "vast-triposg", "name": "TripoSG", "level": "model", "family_id": null, "developer": [ "vast" ], "model_class": "3D", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "VAST open image-to-3D model.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/models?pipeline_tag=image-to-3d&sort=trending", "https://investgame.net/news/matrix-partners-china-has-led-a-446m-series-b-and-b-into-vast-maker-of-tripo-ai-in-its-fourth-round-of-2026/" ], "unverified": [] }, { "id": "vencortex-deosalphatimegptpredictor-2025", "name": "DeOSAlphaTimeGPTPredictor-2025", "level": "model", "family_id": null, "developer": [ "vencortex" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "GIFT-Eval entry from vencortex; its relation to Nixtla TimeGPT is unverified and it is not a TimeGPT score.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/spaces/Salesforce/GIFT-Eval" ], "unverified": [] }, { "id": "vicuna-7b-v1-5", "name": "Vicuna-7B-v1.5", "level": "model", "family_id": null, "developer": [ "lmsys" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": 7000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Language model used inside MAIRA-2; developer not stated in these sections.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/microsoft/maira-2", "https://huggingface.co/lmsys/vicuna-7b-v1.5" ], "unverified": [] }, { "id": "wav2letter", "name": "Wav2letter", "level": "model", "family_id": null, "developer": [], "model_class": "TinyML", "modalities_in": [ "text" ], "modalities_out": [ "labels and boxes" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "microcontroller", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Model listed in Arm ML-Zoo, optimised for Arm IP (Cortex-A, Cortex-M, Mali, Ethos-U) in TensorFlow Lite format.", "verified_on": "2026-10-10", "sources": [ "https://github.com/ARM-software/ML-zoo" ], "unverified": [] }, { "id": "waymo-world-model", "name": "Waymo World Model", "level": "model", "family_id": null, "developer": [ "waymo" ], "model_class": "driving", "modalities_in": [ "video" ], "modalities_out": [ "video", "sensor" ], "architecture": null, "architecture_detail": "Genie 3 variant; takes dashcam or phone video; lidar output; controls for driving actions, layout, language", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-02", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "internal only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Waymo driving simulator built on Genie 3, supporting Waymo's expansion to 11 US cities.", "verified_on": "2026-10-10", "sources": [ "https://en.wikipedia.org/wiki/Genie_(AI_model)" ], "unverified": [] }, { "id": "wayve-gaia", "name": "GAIA", "level": "family", "family_id": null, "developer": [ "wayve" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Wayve generative world models for driving.", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/gaia_3_wayve" ], "unverified": [] }, { "id": "wayve-gaia-1", "name": "GAIA-1", "level": "model", "family_id": "wayve-gaia", "developer": [ "wayve" ], "model_class": "driving", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": 9000000000, "params_active": null, "params_checkpoint": null, "params_note": "About 9B", "context_length": null, "release_date": "2023-06", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First Wayve GAIA world model (Jun/Oct 2023).", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/gaia_3_wayve" ], "unverified": [] }, { "id": "wayve-gaia-2", "name": "GAIA-2", "level": "model", "family_id": "wayve-gaia", "developer": [ "wayve" ], "model_class": "driving", "modalities_in": [], "modalities_out": [], "architecture": "diffusion", "architecture_detail": "Multi-view latent diffusion", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03-26", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Wayve world model, Mar 2025.", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/gaia_3_wayve" ], "unverified": [] }, { "id": "wayve-gaia-3", "name": "GAIA-3", "level": "model", "family_id": "wayve-gaia", "developer": [ "wayve" ], "model_class": "driving", "modalities_in": [ "video", "action" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": "Latent diffusion; video tokenizer 2x GAIA-2's; 10x GAIA-2 data across 9 countries", "params_total": 15000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-12-02", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Wayve driving world model offered for autonomous driving validation; vendor says synthetic-test rejection rates fell fivefold.", "verified_on": "2026-10-10", "sources": [ "https://aiwiki.ai/wiki/gaia_3_wayve", "https://www.autonomousvehicleinternational.com/news/ai-sensor-fusion/wayves-gaia-3-generative-world-model-now-available-for-autonomous-driving-validation.html" ], "unverified": [ "access_routes" ] }, { "id": "weathermesh4", "name": "WeatherMesh4", "level": "model", "family_id": null, "developer": [], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Weather model on WeatherBench 2 (2020).", "verified_on": "2026-10-10", "sources": [ "https://researchsites.withgoogle.com/weatherbench/deterministic/" ], "unverified": [] }, { "id": "wonderworld", "name": "WonderWorld", "level": "model", "family_id": null, "developer": [ "stanford-university" ], "model_class": "world model", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-06-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Model scored in the WorldScore paper table.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/html/2504.00983v2", "https://arxiv.org/html/2406.09394", "https://arxiv.org/abs/2406.09394" ], "unverified": [] }, { "id": "world-labs-marble", "name": "Marble", "level": "family", "family_id": null, "developer": [ "world-labs" ], "model_class": "world model", "modalities_in": [], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-11-12", "release_date_basis": "secondary report", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "World Labs 3D world generation.", "verified_on": "2026-10-10", "sources": [ "https://worldlabs.ai/blog/bigger-better-worlds", "https://siliconangle.com/2025/11/12/world-labs-launches-marble-commercial-world-model-generating-entire-virtual-environments/", "https://www.worldlabs.ai/blog/marble-world-model" ], "unverified": [] }, { "id": "world-labs-marble-1-0", "name": "Marble 1.0", "level": "model", "family_id": "world-labs-marble", "developer": [ "world-labs" ], "model_class": "world model", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": "Multimodal input (text, image, multi-image, 360 panorama, video) to Gaussian-splat worlds (.spz, .ply)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-11", "release_date_basis": "secondary report", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "World Labs 3D world generator; World API added Jan 2026 (also 1.0 Draft).", "verified_on": "2026-10-10", "sources": [ "https://worldlabs.ai/blog/bigger-better-worlds", "https://blog.buildfastwithai.com/tencent-alibaba-world-models-april-2026" ], "unverified": [] }, { "id": "world-labs-marble-1-1", "name": "Marble 1.1", "level": "model", "family_id": "world-labs-marble", "developer": [ "world-labs" ], "model_class": "world model", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Marble update; 1,500 credits per world. Japanese coverage dates the announcement 8 Apr.", "verified_on": "2026-10-10", "sources": [ "https://worldlabs.ai/blog/bigger-better-worlds", "https://www.moguravr.com/?p=307971" ], "unverified": [] }, { "id": "world-labs-marble-1-1-plus", "name": "Marble 1.1 Plus", "level": "variant", "family_id": "world-labs-marble-1-1", "developer": [ "world-labs" ], "model_class": "world model", "modalities_in": [], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "consumer app", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Adds dynamic cubes, up to 5 extra spatial units per generation at +300 credits each.", "verified_on": "2026-10-10", "sources": [ "https://worldlabs.ai/blog/bigger-better-worlds", "https://www.moguravr.com/?p=307971" ], "unverified": [] }, { "id": "xenova-all-minilm-l6-v2", "name": "all-MiniLM-L6-v2 (Xenova, Transformers.js ONNX)", "level": "variant", "family_id": "sentence-transformers-all-minilm-l6-v2", "developer": [ "xenova" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "ONNX export of sentence-transformers/all-MiniLM-L6-v2 for Transformers.js", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-05-02", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "laptop", "offline capable" ], "footprint_note": "Runs in JavaScript through Transformers.js with ONNX weights", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ONNX version of all-MiniLM-L6-v2 for in-browser and JavaScript sentence embeddings through Transformers.js.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/Xenova/all-MiniLM-L6-v2", "https://huggingface.co/api/models/Xenova/all-MiniLM-L6-v2" ], "unverified": [ "deployment_tiers" ] }, { "id": "xiaomi-mimo", "name": "MiMo", "level": "family", "family_id": null, "developer": [ "xiaomi" ], "model_class": "reasoning", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Xiaomi MiMo model line, including the MiMo-V2, V2.5 and V2.6 series.", "verified_on": "2026-10-11", "sources": [ "https://mimo.xiaomi.com/mimo-v2-5", "https://huggingface.co/XiaomiMiMo/MiMo-V2.5" ], "unverified": [] }, { "id": "xiaomi-mimo-v2-5", "name": "MiMo-V2.5", "level": "model", "family_id": "xiaomi-mimo", "developer": [ "xiaomi" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video", "audio" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "MiMo-V2-Flash hybrid sliding-window attention backbone with a 729M-param ViT and a 261M-param audio encoder", "params_total": 310000000000, "params_active": 15000000000, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": "2026-04-22", "release_date_basis": "announcement page", "weights_date": "2026-04-27", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "API", "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Hugging Face language tags: en, zh", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Xiaomi native omnimodal agentic model that understands text, image, video and audio, with up to one million tokens of context.", "verified_on": "2026-10-11", "sources": [ "https://mimo.xiaomi.com/mimo-v2-5", "https://huggingface.co/XiaomiMiMo/MiMo-V2.5", "https://artificialanalysis.ai/models/mimo-v2-5-0424" ], "unverified": [] }, { "id": "xiaomi-mimo-v2-5-pro", "name": "MiMo-V2.5-Pro", "level": "model", "family_id": null, "developer": [ "xiaomi" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04-27", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://arena.ai/leaderboard/chat/text", "https://huggingface.co/XiaomiMiMo/MiMo-V2.5-Pro" ], "unverified": [] }, { "id": "xiaomi-mimo-v2-6-flash", "name": "MiMo-V2.6-Flash", "level": "model", "family_id": "xiaomi-mimo", "developer": [ "xiaomi" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video", "audio" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "Hybrid sliding-window and global attention backbone, 256 routed experts with 8 active, 681M-param ViT and audio encoders", "params_total": 309000000000, "params_active": 15000000000, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": "2026-09-21", "release_date_basis": "registry date", "weights_date": "2026-09-21", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": "Released as Flash-RL and later Flash-MOPD checkpoints; MOPD targets tool-call repetition", "runtimes": [ "sglang", "vllm" ], "languages_note": "Hugging Face language tags: en, zh", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Xiaomi omnimodal MoE model for long-horizon agentic coding and tool use with a one-million-token context.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL", "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-MOPD", "https://artificialanalysis.ai/models/mimo-v2-6-flash" ], "unverified": [] }, { "id": "xiaomi-mimo-v2-6-pro", "name": "MiMo-V2.6-Pro", "level": "model", "family_id": null, "developer": [ "xiaomi" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video", "audio" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "Sparse MoE (384 experts, 8 active, no shared experts), 60 SWA + 10 global layers, 5-layer MTP", "params_total": 1020000000000, "params_active": 42000000000, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": "2026-09", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "multi-GPU node" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Xiaomi open model; highest-scoring open-weights model on AA.", "verified_on": "2026-10-10", "sources": [ "https://artificialanalysis.ai/leaderboards/models", "https://arena.ai/leaderboard/chat/text", "https://artificialanalysis.ai/models/open-source", "https://artificialanalysis.ai/leaderboards/models?open_weights=true", "https://huggingface.co/XiaomiMiMo", "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-RL" ], "unverified": [ "release_date" ] }, { "id": "xiaomi-mimo-v2-6-pro-distilled-qwen-9b", "name": "MiMo-V2.6-Pro distilled Qwen-9B variant", "level": "variant", "family_id": null, "developer": [ "xiaomi" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 9000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "A distilled Qwen-9B variant of MiMo-V2.6-Pro (name not given in the notes).", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/XiaomiMiMo", "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-RL" ], "unverified": [] }, { "id": "xiaomi-mimo-v2-6-pro-mopd", "name": "MiMo-V2.6-Pro-MOPD", "level": "variant", "family_id": null, "developer": [ "xiaomi" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MOPD variant of MiMo-V2.6-Pro.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/XiaomiMiMo", "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-RL" ], "unverified": [] }, { "id": "xiaomi-mimo-v2-6-pro-rl", "name": "MiMo-V2.6-Pro-RL", "level": "variant", "family_id": null, "developer": [ "xiaomi" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "RL variant of MiMo-V2.6-Pro.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/XiaomiMiMo", "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-RL" ], "unverified": [] }, { "id": "xiaomi-mimo-v2-flash", "name": "MiMo-V2-Flash", "level": "model", "family_id": "xiaomi-mimo", "developer": [ "xiaomi" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "Hybrid attention interleaving sliding-window (128-token window) and global attention at 5:1, with multi-token prediction", "params_total": 309000000000, "params_active": 15000000000, "params_checkpoint": null, "params_note": null, "context_length": 256000, "release_date": "2025-12-16", "release_date_basis": "announcement page", "weights_date": "2025-12-16", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "API", "open weights download" ], "deployment_tiers": [], "footprint_note": "Trained on 27T tokens with FP8 mixed precision. Artificial Analysis tracks a February 2026 version as MiMo-V2-Flash (Feb 2026)", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Xiaomi MoE language model built for fast reasoning, coding and agentic workflows, with a hybrid thinking mode and a 256K context window.", "verified_on": "2026-10-11", "sources": [ "https://mimo.xiaomi.com/mimo-v2-flash", "https://huggingface.co/XiaomiMiMo/MiMo-V2-Flash", "https://artificialanalysis.ai/models/mimo-v2-0206" ], "unverified": [] }, { "id": "xiaomi-mimo-v2-omni", "name": "MiMo-V2-Omni", "level": "model", "family_id": "xiaomi-mimo", "developer": [ "xiaomi" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video", "audio" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 256000, "release_date": "2026-03-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "Artificial Analysis tracks the 0327 version as MiMo-V2-Omni-0327 and lists it as proprietary", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Xiaomi omni foundation model that fuses image, video and audio encoders into one backbone, with native tool calling and UI grounding for agents.", "verified_on": "2026-10-11", "sources": [ "https://mimo.xiaomi.com/mimo-v2-omni", "https://artificialanalysis.ai/models/mimo-v2-omni-0327" ], "unverified": [] }, { "id": "xiaomi-mimo-v2-pro", "name": "MiMo-V2-Pro", "level": "model", "family_id": null, "developer": [ "xiaomi" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://leaderboard.steel.dev/leaderboards/swe-bench-verified.md" ], "unverified": [] }, { "id": "yandex-catboost", "name": "CatBoost", "level": "model", "family_id": null, "developer": [ "yandex" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "gradient boosting", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-02-18", "release_date_basis": "registry date", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "1.2.10 uploaded 18 Feb 2026 on PyPI (GitHub 19 Feb)", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Gradient boosting library; strongest GBDT out of the box on TabArena.", "verified_on": "2026-10-10", "sources": [ "https://github.com/catboost/catboost/releases", "https://pypi.org/pypi/catboost/json", "https://github.com/catboost/catboost" ], "unverified": [] }, { "id": "yellow-ai-komodo-7b", "name": "Komodo-7B", "level": "model", "family_id": null, "developer": [ "yellow-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 7000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-03-14", "release_date_basis": "notes", "weights_date": null, "openness": "open weights restricted", "licence": "llama-2-community", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": "English, Indonesian + 11 regional languages", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Yellow.ai's Indonesian 7B model (base and instruct) with expanded vocabulary (35,008 tokens). Komodo-7B-Instruct reports an average of 71.1.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/Yellow-AI-NLP/komodo-7b-base" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "yhat-labs-chakrats", "name": "ChakraTS", "level": "model", "family_id": null, "developer": [ "yhat-labs" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "YHat Labs pretrained forecasting model on GIFT-Eval.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/spaces/Salesforce/GIFT-Eval" ], "unverified": [] }, { "id": "yolo-v3-tiny", "name": "YOLO v3 Tiny", "level": "model", "family_id": null, "developer": [], "model_class": "TinyML", "modalities_in": [ "text" ], "modalities_out": [ "labels and boxes" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "microcontroller", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Model listed in Arm ML-Zoo, optimised for Arm IP (Cortex-A, Cortex-M, Mali, Ethos-U) in TensorFlow Lite format.", "verified_on": "2026-10-10", "sources": [ "https://github.com/ARM-software/ML-zoo" ], "unverified": [] }, { "id": "zhipu-zai-glm", "name": "GLM", "level": "family", "family_id": null, "developer": [ "zhipu-zai" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://docs.z.ai/guides/overview/pricing", "https://huggingface.co/zai-org" ], "unverified": [] }, { "id": "zhipu-zai-glm-4-0520", "name": "GLM-4 (0520)", "level": "model", "family_id": "zhipu-zai-glm", "developer": [ "zhipu-zai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 128000, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Pre-trained on ten trillion tokens, mostly Chinese and English, plus a small corpus in 24 languages; aligned mainly for Chinese and English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "GLM-4 (0520) is the upgraded GLM-4 version reported in the ChatGLM technical report, which says it rivals GPT-4 on MMLU, GSM8K, MATH, BBH, GPQA and HumanEval.", "verified_on": "2026-10-11", "sources": [ "https://arxiv.org/abs/2406.12793", "https://arxiv.org/html/2406.12793v2" ], "unverified": [ "context_length", "access_routes", "openness" ] }, { "id": "zhipu-zai-glm-4-5", "name": "GLM-4.5", "level": "model", "family_id": "zhipu-zai-glm", "developer": [ "zhipu-zai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 355000000000, "params_active": 32000000000, "params_checkpoint": null, "params_note": "23T tokens; GPU count and cost not published.", "context_length": null, "release_date": "2025-08-08", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Zhipu (Z.ai) MoE model; no training cost published.", "verified_on": "2026-10-10", "sources": [ "https://docs.z.ai/guides/overview/pricing", "https://arxiv.org/abs/2508.06471" ], "unverified": [] }, { "id": "zhipu-zai-glm-4-5-air-base", "name": "GLM-4.5-Air-Base", "level": "model", "family_id": "zhipu-zai-glm", "developer": [ "zhipu-zai" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Base text model named as the Hugging Face base_model of GLM-4.5V.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/zai-org/GLM-4.5V" ], "unverified": [] }, { "id": "zhipu-zai-glm-4-5v", "name": "GLM-4.5V", "level": "model", "family_id": "zhipu-zai-glm", "developer": [ "zhipu-zai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "Built on the GLM-4.5-Air text foundation model; thinking mode switch", "params_total": 106000000000, "params_active": 12000000000, "params_checkpoint": null, "params_note": null, "context_length": 65536, "release_date": "2025-08-11", "release_date_basis": "announcement page", "weights_date": "2025-08-10", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "API", "open weights download" ], "deployment_tiers": [], "footprint_note": "16K max output tokens on the Z.ai API", "runtimes": [ "sglang" ], "languages_note": "Hugging Face language tags: zh, en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Z.ai vision-language foundation model for multimodal agent applications covering image, video, document and GUI tasks.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/zai-org/GLM-4.5V", "https://docs.z.ai/guides/vlm/glm-4.5v", "https://openrouter.ai/z-ai/glm-4.5v" ], "unverified": [] }, { "id": "zhipu-zai-glm-4-6", "name": "GLM-4.6", "level": "model", "family_id": "zhipu-zai-glm", "developer": [ "zhipu-zai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://docs.z.ai/guides/overview/pricing" ], "unverified": [] }, { "id": "zhipu-zai-glm-4-6v", "name": "GLM-4.6V", "level": "model", "family_id": "zhipu-zai-glm", "developer": [ "zhipu-zai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": 106000000000, "params_active": null, "params_checkpoint": null, "params_note": "Series also includes GLM-4.6V-Flash (9B) for local deployment", "context_length": 128000, "release_date": "2025-12-08", "release_date_basis": "registry date", "weights_date": "2025-12-07", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "API", "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Hugging Face language tags: zh, en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Z.ai vision-language model for visual understanding and long-context reasoning across images, documents and mixed media, with native multimodal function calling.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/zai-org/GLM-4.6V", "https://docs.z.ai/guides/vlm/glm-4.6v", "https://openrouter.ai/z-ai/glm-4.6v" ], "unverified": [] }, { "id": "zhipu-zai-glm-4-7", "name": "GLM-4.7", "level": "model", "family_id": "zhipu-zai-glm", "developer": [ "zhipu-zai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://docs.z.ai/guides/overview/pricing" ], "unverified": [] }, { "id": "zhipu-zai-glm-4-7-flash", "name": "GLM 4.7 Flash", "level": "model", "family_id": null, "developer": [ "zhipu-zai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01-19", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "measured on a benchmark", "regional_languages": [], "indonesia_hosting": "in-region Jakarta", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://leaderboard.sea-lion.ai/detailed/ID", "https://huggingface.co/zai-org/GLM-4.7-Flash", "https://docs.aws.amazon.com/bedrock/latest/userguide/models-region-compatibility.md" ], "unverified": [], "hosting_provider": "Amazon Bedrock (ap-southeast-3 Jakarta)", "hosting_url": "https://docs.aws.amazon.com/bedrock/latest/userguide/models-region-compatibility.md", "indonesian_measured": [ { "benchmark": "SEA-HELM Indonesian leaderboard (ID aggregate, first score column of the Indonesian Competency and Task Scores table)", "score": "67.89 (95% CI -1.32 to +1.27)", "metric": "score 0 to 100, average of 2000 bootstraps with 95% CI", "run_by": "AI Singapore (SEA-HELM leaderboard maintainer)", "url": "https://leaderboard.sea-lion.ai/detailed/ID" } ] }, { "id": "zhipu-zai-glm-4-9b-0414", "name": "GLM-4-9B-0414", "level": "model", "family_id": "zhipu-zai-glm", "developer": [ "zhipu-zai" ], "model_class": "LLM", "modalities_in": [], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Z.ai 9B text model used to initialize the autoregressive generator of GLM-Image.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/zai-org/GLM-Image" ], "unverified": [] }, { "id": "zhipu-zai-glm-5", "name": "GLM-5", "level": "model", "family_id": "zhipu-zai-glm", "developer": [ "zhipu-zai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://docs.z.ai/guides/overview/pricing" ], "unverified": [] }, { "id": "zhipu-zai-glm-5-1", "name": "GLM-5.1", "level": "model", "family_id": "zhipu-zai-glm", "developer": [ "zhipu-zai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04-03", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": null, "verified_on": "2026-10-10", "sources": [ "https://docs.z.ai/guides/overview/pricing", "https://arena.ai/leaderboard/chat/text", "https://huggingface.co/zai-org/GLM-5.1" ], "unverified": [] }, { "id": "zhipu-zai-glm-5-2", "name": "GLM-5.2", "level": "model", "family_id": "zhipu-zai-glm", "developer": [ "zhipu-zai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 753000000000, "params_active": 40000000000, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": "2026-06-13", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "hyperscale cluster", "multi-GPU node" ], "footprint_note": "T2/T3 estimate (unverified). About 40B active.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Z.ai open-weight coder-strong MoE model.", "verified_on": "2026-10-10", "sources": [ "https://rits.shanghai.nyu.edu/ai/glm-5-2-z-ais-open-weights-coder-beats-gpt-5-5-at-1-6-the-cost/", "https://leaderboard.steel.dev/leaderboards/swe-bench-verified.md", "https://huggingface.co/zai-org/GLM-5.3", "https://arena.ai/leaderboard/chat/text", "https://www.benchlm.ai/benchmarks/tau2-bench" ], "unverified": [ "deployment_tiers" ] }, { "id": "zhipu-zai-glm-5-3", "name": "GLM-5.3", "level": "model", "family_id": "zhipu-zai-glm", "developer": [ "zhipu-zai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "MoE with DSA attention (tag glm_moe_dsa); same base as GLM-5.2", "params_total": 753000000000, "params_active": null, "params_checkpoint": null, "params_note": "Active parameters unverified; context: evaluated up to 1M", "context_length": 1000000, "release_date": "2026-08-25", "release_date_basis": "secondary report", "weights_date": "2026-09-04", "openness": "open weights restricted", "licence": "glm-5-3-license", "access_routes": [ "API", "open weights download" ], "deployment_tiers": [ "hyperscale cluster", "multi-GPU node" ], "footprint_note": "T2/T3 estimate (unverified). Active parameters not stated. Up to 1M context in evals.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Z.ai's open-weight GLM-5 line model; one of the highest-scoring open models on AA. Licence: custom file, not MIT as Arena labels it.", "verified_on": "2026-10-10", "sources": [ "https://huggingface.co/zai-org/GLM-5.3", "https://huggingface.co/zai-org", "https://docs.z.ai/guides/overview/pricing", "https://artificialanalysis.ai/leaderboards/models", "https://arena.ai/leaderboard/chat/text", "https://artificialanalysis.ai/models/open-source", "https://huggingface.co/zai-org/GLM-5.3/blob/main/LICENSE", "https://codersera.com/blog/open-source-llms-landscape-2026/", "https://artificialanalysis.ai/leaderboards/models?open_weights=true" ], "unverified": [ "deployment_tiers", "params_active" ] }, { "id": "zhipu-zai-glm-5-3-flash", "name": "GLM-5.3-Flash", "level": "model", "family_id": "zhipu-zai-glm", "developer": [ "zhipu-zai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "New base model; hybrid sparse + linear attention, mHC; 30T-token multimodal pretraining corpus", "params_total": 320000000000, "params_active": 18000000000, "params_checkpoint": null, "params_note": "Context evaluated up to 1M", "context_length": 1000000, "release_date": "2026-08-25", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "API", "open weights download" ], "deployment_tiers": [ "multi-GPU node" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Z.ai's 320B MIT-licensed model on a new base. Adapters 3, finetunes 27, merges 4, quants 165.", "verified_on": "2026-10-10", "sources": [ "https://docs.z.ai/guides/overview/pricing", "https://huggingface.co/zai-org", "https://codersera.com/blog/open-source-llms-landscape-2026/", "https://huggingface.co/zai-org/GLM-5.3-Flash", "https://arena.ai/leaderboard/chat/text" ], "unverified": [] }, { "id": "zhipu-zai-glm-5-3-flashx", "name": "GLM-5.3-FlashX", "level": "model", "family_id": "zhipu-zai-glm", "developer": [ "zhipu-zai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Served commercially without weights.", "verified_on": "2026-10-10", "sources": [ "https://docs.z.ai/guides/overview/pricing", "https://codersera.com/blog/open-source-llms-landscape-2026/" ], "unverified": [] }, { "id": "zhipu-zai-glm-5-3-prime", "name": "GLM-5.3-Prime", "level": "model", "family_id": null, "developer": [ "zhipu-zai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Served commercially without weights.", "verified_on": "2026-10-10", "sources": [ "https://codersera.com/blog/open-source-llms-landscape-2026/" ], "unverified": [] }, { "id": "zhipu-zai-glm-5v-turbo", "name": "GLM-5V-Turbo", "level": "model", "family_id": "zhipu-zai-glm", "developer": [ "zhipu-zai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 200000, "release_date": "2026-04-01", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "128K max output tokens; thinking mode, function calling and context caching", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Z.ai first native multimodal agent foundation model, built for vision-based coding and agent workflows over image, video and text input.", "verified_on": "2026-10-11", "sources": [ "https://docs.z.ai/guides/vlm/glm-5v-turbo", "https://openrouter.ai/z-ai/glm-5v-turbo" ], "unverified": [] }, { "id": "zhipu-zai-glm-asr-nano", "name": "GLM-ASR-Nano (GLM-ASR-Nano-2512)", "level": "model", "family_id": null, "developer": [ "zhipu-zai" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "autoregressive", "architecture_detail": "Llama-type text decoder (about 1.59B) + audio encoder (about 0.63B) + projector, per header reading", "params_total": 1500000000, "params_active": null, "params_checkpoint": 2260000000, "params_note": "1.5B is Z.ai's stated figure; 2.26B (2,257,843,200) is the stored checkpoint; what 1.5B counts is an open question", "context_length": null, "release_date": "2025-12-09", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": "17 languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Z.ai open speech recognition model.", "verified_on": "2026-10-10", "sources": [ "https://github.com/modelscope/FunASR", "https://huggingface.co/api/models/zai-org/GLM-ASR-Nano-2512", "https://huggingface.co/zai-org/GLM-ASR-Nano-2512", "https://github.com/zai-org/GLM-ASR" ], "unverified": [] }, { "id": "zhipu-zai-glm-image", "name": "GLM-Image", "level": "model", "family_id": "zhipu-zai-glm", "developer": [ "zhipu-zai" ], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": "hybrid", "architecture_detail": "Autoregressive generator (9B, initialized from GLM-4-9B-0414) plus a 7B single-stream DiT diffusion decoder", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "9B autoregressive generator plus 7B diffusion decoder", "context_length": null, "release_date": "2026-01", "release_date_basis": "secondary report", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": "VQ tokenizer and ViT weights from X-Omni remain under Apache 2.0", "runtimes": [ "sglang" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Z.ai image generation model with a hybrid autoregressive and diffusion design, strong at text rendering and knowledge-intensive generation, and also supporting image editing.", "verified_on": "2026-10-11", "sources": [ "https://huggingface.co/zai-org/GLM-Image", "https://artificialanalysis.ai/image/leaderboard/text-to-image" ], "unverified": [] }, { "id": "zhipu-zai-glm-ocr", "name": "GLM-OCR", "level": "model", "family_id": null, "developer": [ "zhipu-zai" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "autoregressive", "architecture_detail": "0.4B CogViT visual encoder + 0.5B GLM language decoder; checkpoint also holds an untied output head and one MTP layer", "params_total": 900000000, "params_active": null, "params_checkpoint": 1330000000, "params_note": "0.9B is Z.ai's official figure; 1.33B (1,325,258,240) is the full stored checkpoint", "context_length": null, "release_date": "2026-01-30", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "single datacentre GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Z.ai compact multimodal OCR model in the GLM family.", "verified_on": "2026-10-10", "sources": [ "https://arxiv.org/pdf/2606.03264", "https://huggingface.co/api/models/zai-org/GLM-OCR", "https://huggingface.co/zai-org/GLM-OCR", "https://arxiv.org/abs/2603.10910" ], "unverified": [] }, { "id": "alexnet", "name": "AlexNet", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Single-GPU variant described in arXiv 1404.5997 (\"One weird trick\"); TorchVision ships the ImageNet weights.", "context_length": null, "release_date": "2012", "release_date_basis": "inferred from page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Deep convolutional image classifier by Krizhevsky, Sutskever and Hinton (NeurIPS 2012) that won ImageNet 2012; still shipped as a baseline in TorchVision, ONNX and cloud model zoos.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1404.5997", "https://papers.nips.cc/paper/2012/hash/c399862d3b9d6b76c8436e924a68c45b-Abstract.html", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://pytorch.org/hub/pytorch_vision_alexnet", "https://raw.githubusercontent.com/pytorch/hub/master/pytorch_vision_alexnet.md", "https://github.com/pytorch/vision/blob/main/torchvision/models/alexnet.py", "https://pytorch.org/vision/stable/models/alexnet.html" ], "unverified": [] }, { "id": "zfnet", "name": "ZFNet", "level": "model", "family_id": null, "developer": [ "nyu" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2013-11-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Image classification CNN from the paper \"Visualizing and Understanding Convolutional Networks\" (Zeiler and Fergus); the ONNX Model Zoo ships ZFNet-512.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1311.2901", "https://github.com/onnx/models/tree/main/validated/vision/classification/zfnet-512", "https://arxiv.org/html/1311.2901" ], "unverified": [] }, { "id": "caffenet", "name": "CaffeNet", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "AlexNet-style ImageNet classifier from the Caffe framework, kept in the ONNX Model Zoo as a reference CNN.", "verified_on": "2026-10-12", "sources": [ "https://github.com/onnx/models/tree/main/validated/vision/classification/caffenet" ], "unverified": [] }, { "id": "vgg", "name": "VGG", "level": "family", "family_id": null, "developer": [ "university-of-oxford" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "VGG-11, VGG-13, VGG-16 and VGG-19, with and without batch normalisation (TorchVision, timm, Keras, ONNX).", "context_length": null, "release_date": "2014-09-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Very deep plain convolutional networks (11 to 19 layers) from the University of Oxford; widely used as an image classifier and as a feature extractor for transfer learning.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1409.1556", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://huggingface.co/timm/vgg16.tv_in1k", "https://www.modelscope.cn/models/facebook/VGGT-1B", "https://pytorch.org/hub/pytorch_vision_vgg", "https://raw.githubusercontent.com/pytorch/hub/master/pytorch_vision_vgg.md", "https://github.com/pytorch/vision/blob/main/torchvision/models/vgg.py", "https://arxiv.org/html/1409.1556" ], "unverified": [] }, { "id": "vgg-16", "name": "VGG-16", "level": "variant", "family_id": "vgg", "developer": [ "university-of-oxford" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": 138000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2014-09-04", "release_date_basis": "announcement page", "weights_date": "2023-04-25", "openness": "open weights permissive", "licence": "bsd-3-clause", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "16-layer VGG image classifier trained on ImageNet; the most used VGG variant.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1409.1556", "https://huggingface.co/timm/vgg16.tv_in1k", "https://arxiv.org/html/1409.1556" ], "unverified": [] }, { "id": "vgg-19", "name": "VGG-19", "level": "variant", "family_id": "vgg", "developer": [ "university-of-oxford" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2014-09-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "19-layer VGG image classifier, the largest VGG configuration in the paper.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1409.1556", "https://arxiv.org/html/1409.1556" ], "unverified": [] }, { "id": "google-inception", "name": "Inception (GoogLeNet)", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Inception v1 (GoogLeNet), v2, v3, v4 and Inception-ResNet v2.", "context_length": null, "release_date": "2014-09-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "mobile or NPU", "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, OpenVINO, TI edge AI model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google family of image classification CNNs built from multi-branch Inception modules, from GoogLeNet (ImageNet 2014) to Inception v4 and Inception-ResNet v2.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1409.4842", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://huggingface.co/timm/inception_v3.tv_in1k", "https://aihub.qualcomm.com/models/googlenet", "https://github.com/pytorch/vision/blob/main/torchvision/models/googlenet.py", "https://aihub.qualcomm.com/models/inception_v3", "https://github.com/pytorch/vision/blob/main/torchvision/models/inception.py" ], "unverified": [] }, { "id": "google-googlenet", "name": "GoogLeNet (Inception v1)", "level": "variant", "family_id": "google-inception", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2014-09-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "The 22-layer Inception v1 network that won the ImageNet 2014 classification challenge.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1409.4842" ], "unverified": [] }, { "id": "google-inception-v3", "name": "Inception v3", "level": "variant", "family_id": "google-inception", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": 23800000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2015-12-02", "release_date_basis": "announcement page", "weights_date": "2023-04-25", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Inception v3 image classifier from the paper \"Rethinking the Inception Architecture for Computer Vision\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1512.00567", "https://huggingface.co/timm/inception_v3.tv_in1k" ], "unverified": [] }, { "id": "google-inception-v4", "name": "Inception v4", "level": "variant", "family_id": "google-inception", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2016-02-23", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Inception v4 image classifier from the paper \"Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1602.07261" ], "unverified": [] }, { "id": "google-inception-resnet-v2", "name": "Inception-ResNet v2", "level": "variant", "family_id": "google-inception", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2016-02-23", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Inception network with residual connections; also the backbone of the Open Images Faster R-CNN detector on Kaggle.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1602.07261" ], "unverified": [] }, { "id": "xception", "name": "Xception", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Keras Applications Xception; timm also ships aligned Xception variants.", "context_length": null, "release_date": "2016-10-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Image classification CNN built on depthwise separable convolutions, by Francois Chollet; part of Keras Applications.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1610.02357", "https://www.kaggle.com/models/keras/xception", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://keras.io/api/applications/", "https://keras.io/keras_hub/presets/", "https://github.com/onnx/models/tree/main/Computer_Vision/xception41_Opset16_timm", "https://arxiv.org/html/1610.02357" ], "unverified": [] }, { "id": "wide-resnet", "name": "Wide ResNet", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": 68951464, "params_active": null, "params_checkpoint": null, "params_note": "WRN-50-2 and WRN-101-2 in TorchVision and timm.", "context_length": null, "release_date": "2016-05-23", "release_date_basis": "announcement page", "weights_date": "2023-04-05", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Residual networks made wider and shallower (Zagoruyko and Komodakis); Wide ResNet-50-2 is a common ImageNet backbone.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1605.07146", "https://huggingface.co/timm/wide_resnet50_2.racm_in1k", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://aihub.qualcomm.com/models/wideresnet50", "https://github.com/pytorch/vision/blob/main/torchvision/models/resnet.py", "https://pytorch.org/hub/pytorch_vision_wide_resnet", "https://raw.githubusercontent.com/pytorch/hub/master/pytorch_vision_wide_resnet.md" ], "unverified": [] }, { "id": "resnext", "name": "ResNeXt", "level": "family", "family_id": null, "developer": [ "uc-san-diego", "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "ResNeXt-50 32x4d, 101 32x8d and 152; Instagram weakly supervised (WSL) and semi-supervised (SSL, SWSL) weights.", "context_length": null, "release_date": "2016-11-16", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Residual network with grouped \"cardinality\" branches, from the paper \"Aggregated Residual Transformations for Deep Neural Networks\"; used as an ImageNet classifier and as a detection backbone in Detectron2.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1611.05431", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://aihub.qualcomm.com/models/resnext50", "https://github.com/pytorch/vision/blob/main/torchvision/models/resnet.py", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/dle/models/resnext_101-32x4d_pyt_ckpt", "https://pytorch.org/hub/pytorch_vision_resnext", "https://raw.githubusercontent.com/pytorch/hub/master/pytorch_vision_resnext.md", "https://arxiv.org/html/1611.05431" ], "unverified": [] }, { "id": "resnext-50-32x4d", "name": "ResNeXt-50 32x4d", "level": "variant", "family_id": "resnext", "developer": [ "uc-san-diego", "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": 25000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2016-11-16", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "The 50-layer ResNeXt with 32 groups of width 4; the standard TorchVision ResNeXt.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1611.05431", "https://arxiv.org/html/1611.05431" ], "unverified": [] }, { "id": "resnext-101-32x8d-wsl", "name": "ResNeXt-101 32x8d WSL", "level": "variant", "family_id": "resnext", "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-05-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "ResNeXt-101 32x8d pretrained on Instagram hashtags (weakly supervised) and fine-tuned on ImageNet.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1611.05431", "https://arxiv.org/html/1611.05431", "https://arxiv.org/abs/1805.00932" ], "unverified": [ "summary" ] }, { "id": "se-resnet", "name": "SE-ResNet and SE-ResNeXt (Squeeze-and-Excitation)", "level": "model", "family_id": null, "developer": [ "momenta", "university-of-oxford" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "SE-ResNet, SE-ResNeXt and SENet-154 (timm, NVIDIA Deep Learning Examples).", "context_length": null, "release_date": "2017-09-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Image classifiers that add Squeeze-and-Excitation channel attention blocks to ResNet and ResNeXt; SENet won ImageNet 2017.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1709.01507", "https://pytorch.org/hub/nvidia_deeplearningexamples_se-resnext", "https://raw.githubusercontent.com/pytorch/hub/master/nvidia_deeplearningexamples_se-resnext.md", "https://github.com/NVIDIA/DeepLearningExamples/tree/master/PyTorch/Classification/ConvNets/se-resnext101-32x4d", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/seresnet", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/onnx/models/tree/main/Computer_Vision/gluon_senet154_Opset16_timm", "https://arxiv.org/html/1709.01507" ], "unverified": [] }, { "id": "res2net", "name": "Res2Net", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-04-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multi-scale residual backbone from the paper \"Res2Net: A New Multi-scale Backbone Architecture\", used for classification and detection.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1904.01169", "https://github.com/open-mmlab/mmdetection/tree/main/configs/res2net", "https://github.com/Res2Net/Res2Net-detectron2", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/onnx/models/tree/main/Computer_Vision/res2net101_26w_4s_Opset16_timm" ], "unverified": [] }, { "id": "resnest", "name": "ResNeSt", "level": "model", "family_id": null, "developer": [ "meta", "amazon" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "ResNeSt-50, 101, 200 and 269.", "context_length": null, "release_date": "2020-04-19", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Split-attention residual network from the paper \"ResNeSt: Split-Attention Networks\", used as a classification and detection backbone.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2004.08955", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/resnest-50-pytorch", "https://www.modelscope.cn/models/iic/cv_resnest101_general_recognition", "https://pytorch.org/hub/pytorch_vision_resnest", "https://raw.githubusercontent.com/pytorch/hub/master/pytorch_vision_resnest.md", "https://github.com/zhanghang1989/ResNeSt", "https://github.com/open-mmlab/mmdetection/tree/main/configs/resnest", "https://arxiv.org/html/2004.08955" ], "unverified": [] }, { "id": "resnet-rs", "name": "ResNet-RS", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-03-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Re-scaled ResNets with improved training from the paper \"Revisiting ResNets: Improved Training and Scaling Strategies\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.07579", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/tensorflow/models/blob/master/official/README.md", "https://github.com/onnx/models/tree/main/Computer_Vision/resnetrs101_Opset16_timm", "https://arxiv.org/html/2103.07579" ], "unverified": [] }, { "id": "densenet", "name": "DenseNet", "level": "model", "family_id": null, "developer": [ "cornell-university", "tsinghua-university", "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": 8062504, "params_active": null, "params_checkpoint": null, "params_note": "DenseNet-121, 161, 169 and 201.", "context_length": null, "release_date": "2016-08-25", "release_date_basis": "announcement page", "weights_date": "2023-04-21", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "mobile or NPU", "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Densely connected convolutional networks in which each layer feeds every later layer; DenseNet-121 is a common medical-imaging backbone (for example CheXNet).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1608.06993", "https://huggingface.co/timm/densenet121.ra_in1k", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://aihub.qualcomm.com/models/densenet121", "https://github.com/pytorch/vision/blob/main/torchvision/models/densenet.py", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/densenet-121-tf", "https://pytorch.org/hub/pytorch_vision_densenet", "https://arxiv.org/html/1608.06993" ], "unverified": [] }, { "id": "chexnet", "name": "CheXNet", "level": "model", "family_id": null, "developer": [ "stanford-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-11-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DenseNet-121 chest X-ray classifier for pneumonia detection from the paper \"CheXNet: Radiologist-Level Pneumonia Detection on Chest X-Rays with Deep Learning\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1711.05225", "https://paperswithcode.com/method/chexnet", "https://huggingface.co/datasets/pwc-archive/methods", "http://arxiv.org/abs/1711.05225v3", "https://arxiv.org/html/1711.05225" ], "unverified": [] }, { "id": "dla", "name": "DLA (Deep Layer Aggregation)", "level": "model", "family_id": null, "developer": [ "uc-berkeley" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": 26401048, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-07-20", "release_date_basis": "announcement page", "weights_date": "2023-04-24", "openness": "open weights permissive", "licence": "bsd-3-clause", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU", "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Backbone that aggregates features across layers, from the paper \"Deep Layer Aggregation\"; DLA-34 is the CenterNet backbone.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1707.06484", "https://huggingface.co/timm/dla102x.in1k", "https://aihub.qualcomm.com/models/dla102x", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/dla-34", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/onnx/models/tree/main/Computer_Vision/dla102_Opset16_timm", "https://arxiv.org/html/1707.06484" ], "unverified": [] }, { "id": "dpn", "name": "DPN (Dual Path Network)", "level": "model", "family_id": null, "developer": [ "nus" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-07-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Image classifier combining residual and densely connected paths, from the paper \"Dual Path Networks\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1707.01629", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/onnx/models/tree/main/Computer_Vision/dpn107_Opset16_timm", "https://arxiv.org/html/1707.01629" ], "unverified": [] }, { "id": "cspnet", "name": "CSPNet", "level": "model", "family_id": null, "developer": [ "academia-sinica" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "CSPResNet, CSPResNeXt, CSPDarkNet-53 and DarkNet-53 in timm.", "context_length": null, "release_date": "2019-11-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Cross Stage Partial backbone design that cuts computation; CSPDarkNet is the backbone of YOLOv4 and later YOLO models.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1911.11929", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/cspnet", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://keras.io/keras_hub/presets/", "https://www.kaggle.com/models/keras/cspdarknet", "https://github.com/onnx/models/tree/main/Computer_Vision/cs3darknet_focus_l_Opset16_timm", "https://arxiv.org/html/1911.11929" ], "unverified": [] }, { "id": "vovnet", "name": "VoVNet", "level": "model", "family_id": null, "developer": [ "etri" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "VoVNet v1 and v2 (v2 from the CenterMask paper, arXiv 1911.06667).", "context_length": null, "release_date": "2019-04-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Energy and GPU-efficient backbone for real-time object detection built from one-shot aggregation modules.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1904.09730", "https://github.com/youngwanLEE/vovnet-detectron2", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/abs/1911.06667", "https://github.com/onnx/models/tree/main/Computer_Vision/ese_vovnet19b_dw_Opset16_timm", "https://arxiv.org/html/1904.09730" ], "unverified": [] }, { "id": "regnet", "name": "RegNet", "level": "family", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "RegNetX and RegNetY from 200MF to 32GF; RegNetZ (arXiv 2103.06877).", "context_length": null, "release_date": "2020-03-30", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU", "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, OpenVINO, TI edge AI model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Family of image classification networks found by searching network design spaces, from the paper \"Designing Network Design Spaces\"; used as classification and detection backbones.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2003.13678", "https://aihub.qualcomm.com/models/regnet", "https://github.com/pytorch/vision/blob/main/torchvision/models/regnet.py", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/regnetx-3.2gf", "https://github.com/TexasInstruments/edgeai-tensorlab/tree/main/edgeai-modelzoo/models/vision/classification/imagenet1k", "https://paperswithcode.com/method/regnety", "https://huggingface.co/datasets/pwc-archive/methods", "https://arxiv.org/html/2003.13678" ], "unverified": [] }, { "id": "regnetz", "name": "RegNetZ", "level": "variant", "family_id": "regnet", "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-03-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "RegNet variant from the paper \"Fast and Accurate Model Scaling\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.06877", "https://arxiv.org/html/2103.06877" ], "unverified": [] }, { "id": "nfnet", "name": "NFNet", "level": "model", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "NFNet-F0 to F6; NF-ResNet and NF-RegNet.", "context_length": null, "release_date": "2021-02-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Normaliser-free image classification networks from the paper \"High-Performance Large-Scale Image Recognition Without Normalization\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2102.06171", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/nfnet-f0", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/onnx/models/tree/main/Computer_Vision/dm_nfnet_f0_Opset16_timm", "https://arxiv.org/html/2102.06171" ], "unverified": [] }, { "id": "bit", "name": "BiT (Big Transfer)", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-12-24", "release_date_basis": "announcement page", "weights_date": "2022-12-06", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ResNetV2 models pretrained on large image datasets for transfer learning, from the paper \"Big Transfer (BiT): General Visual Representation Learning\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1912.11370", "https://huggingface.co/google/bit-50", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://modelzoo.co/model/big-transfer", "https://github.com/google-research/big_transfer", "https://www.kaggle.com/models/google/bit", "https://www.kaggle.com/models/google/experts-bit" ], "unverified": [] }, { "id": "tresnet", "name": "TResNet", "level": "model", "family_id": null, "developer": [ "alibaba-damo" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-03-30", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "GPU-dedicated high-throughput image classifier from the paper \"TResNet: High Performance GPU-Dedicated Architecture\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2003.13630", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2003.13630" ], "unverified": [] }, { "id": "eca-resnet", "name": "ECA-ResNet", "level": "model", "family_id": null, "developer": [ "tianjin-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-10-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ResNet with Efficient Channel Attention, from the paper \"ECA-Net: Efficient Channel Attention for Deep Convolutional Neural Networks\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1910.03151", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/onnx/models/tree/main/Computer_Vision/eca_botnext26ts_256_Opset16_timm", "https://arxiv.org/html/1910.03151" ], "unverified": [] }, { "id": "sk-resnet", "name": "Selective Kernel Networks (SK-ResNet)", "level": "model", "family_id": null, "developer": [ "njust", "momenta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-03-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ResNet variant that adapts its receptive field with selective kernel attention, from the paper \"Selective Kernel Networks\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1903.06586", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/onnx/models/tree/main/Computer_Vision/skresnet18_Opset16_timm", "https://arxiv.org/html/1903.06586" ], "unverified": [] }, { "id": "ibn-net", "name": "IBN-Net", "level": "model", "family_id": null, "developer": [ "cuhk", "sensetime" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-07-25", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Backbone mixing instance and batch normalisation for better generalisation across domains, from the paper \"Two at Once\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1807.09441", "https://github.com/XingangPan/IBN-Net", "https://pytorch.org/hub/pytorch_vision_ibnnet", "https://raw.githubusercontent.com/pytorch/hub/master/pytorch_vision_ibnnet.md", "https://arxiv.org/html/1807.09441" ], "unverified": [] }, { "id": "hardnet", "name": "HarDNet", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-09-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Low memory traffic CNN backbone for classification and segmentation, from the paper \"HarDNet: A Low Memory Traffic Network\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1909.00948", "https://github.com/PingoLH/Pytorch-HarDNet", "https://pytorch.org/hub/pytorch_vision_hardnet", "https://raw.githubusercontent.com/pytorch/hub/master/pytorch_vision_hardnet.md" ], "unverified": [] }, { "id": "simplenet-classifier", "name": "SimpleNet (classifier)", "level": "model", "family_id": null, "developer": [ "islamic-azad-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2016-08-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Simple plain CNN image classifier from the paper \"Lets keep it simple, Using simple architectures to outperform deeper and more complex architectures\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1608.06037", "https://github.com/Coderx7/SimpleNet_Pytorch", "https://pytorch.org/hub/simplenet", "https://raw.githubusercontent.com/pytorch/hub/master/simplenet.md", "https://github.com/Coderx7/SimpleNet_Pytorch/blob/master/imagenet/simplenet.py", "https://arxiv.org/html/1608.06037" ], "unverified": [] }, { "id": "meal-v2", "name": "MEAL V2", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-09-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ResNet-50 classifiers improved by multi-model ensemble distillation; the MEAL method paper is \"MEAL: Multi-Model Ensemble via Adversarial Learning\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1812.02425", "https://github.com/szq0214/MEAL-V2", "https://pytorch.org/hub/pytorch_vision_meal_v2", "https://raw.githubusercontent.com/pytorch/hub/master/pytorch_vision_meal_v2.md", "https://arxiv.org/abs/2009.08453" ], "unverified": [] }, { "id": "rdnet", "name": "RDNet (DenseNets Reloaded)", "level": "model", "family_id": null, "developer": [ "naver" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-03-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Modernised DenseNet image classifier from the paper \"DenseNets Reloaded: Paradigm Shift Beyond ResNets and ViTs\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2403.19588", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://ar5iv.labs.arxiv.org/html/2403.19588" ], "unverified": [] }, { "id": "repvgg", "name": "RepVGG", "level": "model", "family_id": null, "developer": [ "tsinghua-university", "megvii" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": 9132616, "params_active": null, "params_checkpoint": null, "params_note": "RepVGG A0 to B3.", "context_length": null, "release_date": "2021-01-11", "release_date_basis": "announcement page", "weights_date": "2023-08-23", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "VGG-style plain CNN that is trained with branches and re-parameterised into a single path for fast inference.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2101.03697", "https://huggingface.co/timm/repvgg_a0.rvgg_in1k", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/repvgg-a0", "https://huggingface.co/api/models?pipeline_tag=image-classification&sort=downloads&limit=50", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/repvgg", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/onnx/models/tree/main/Computer_Vision/repvgg_a2_Opset16_timm", "https://arxiv.org/html/2101.03697" ], "unverified": [] }, { "id": "google-mobilenet", "name": "MobileNet", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "MobileNet v1, v2 (already in AURION as mobilenet-v2), v3 (Large, Small), v4, v5 and MobileNet-EdgeTPU.", "context_length": null, "release_date": "2017-04-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "mobile or NPU", "CPU server", "microcontroller" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, OpenVINO, STM32 model zoo, NXP eIQ model zoo, TI edge AI model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google family of efficient CNNs for mobile and embedded vision, used as classifiers and as backbones for SSD, DeepLab and other edge models.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1704.04861", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://huggingface.co/timm/mobilenetv3_small_100.lamb_in1k", "https://huggingface.co/google/mobilenet_v1_1.0_224", "https://aihub.qualcomm.com/models/mobilenet_v3_large", "https://github.com/pytorch/vision/blob/main/torchvision/models/mobilenetv3.py", "https://aihub.qualcomm.com/models/mobilenet_v4" ], "unverified": [] }, { "id": "google-mobilenet-v1", "name": "MobileNet v1", "level": "variant", "family_id": "google-mobilenet", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-04-17", "release_date_basis": "announcement page", "weights_date": "2022-11-10", "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "The first MobileNet, built on depthwise separable convolutions with width and resolution multipliers.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1704.04861", "https://huggingface.co/google/mobilenet_v1_1.0_224" ], "unverified": [] }, { "id": "google-mobilenet-v3", "name": "MobileNet v3", "level": "variant", "family_id": "google-mobilenet", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": 2554968, "params_active": null, "params_checkpoint": null, "params_note": "Large and Small; LR-ASPP segmentation head and SSDLite detector come from the same paper.", "context_length": null, "release_date": "2019-05-06", "release_date_basis": "announcement page", "weights_date": "2022-12-16", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MobileNet found by hardware-aware architecture search, from the paper \"Searching for MobileNetV3\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1905.02244", "https://huggingface.co/timm/mobilenetv3_small_100.lamb_in1k" ], "unverified": [] }, { "id": "google-mobilenet-v4", "name": "MobileNet v4", "level": "variant", "family_id": "google-mobilenet", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-04-16", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "MobileNet generation designed to run efficiently across mobile CPUs, GPUs, DSPs and NPUs, from the paper \"MobileNetV4: Universal Models for the Mobile Ecosystem\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2404.10518" ], "unverified": [] }, { "id": "google-mobilenet-v5", "name": "MobileNet v5", "level": "variant", "family_id": "google-mobilenet", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vision encoder of the Gemma 3n on-device models, described on the Gemma 3n documentation page.", "verified_on": "2026-10-12", "sources": [ "https://ai.google.dev/gemma/docs/gemma-3n", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://keras.io/keras_hub/presets/", "https://www.kaggle.com/models/keras/mobilenetv5" ], "unverified": [ "modalities_out" ] }, { "id": "google-mobilenet-edgetpu", "name": "MobileNet-EdgeTPU", "level": "variant", "family_id": "google-mobilenet", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MobileNet variants optimised for the Google Edge TPU accelerator, published on Kaggle Models.", "verified_on": "2026-10-12", "sources": [ "https://www.kaggle.com/models/google/mobilenet-edgetpu-v2", "https://github.com/tensorflow/tfhub.dev/tree/master/assets/docs/google/collections/mobilenet-edgetpu-v2/1.md", "https://github.com/tensorflow/models/tree/master/official/projects/edgetpu" ], "unverified": [] }, { "id": "mnasnet", "name": "MnasNet", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-07-31", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Mobile image classifier found by platform-aware neural architecture search, from the paper \"MnasNet\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1807.11626", "https://aihub.qualcomm.com/models/mnasnet05", "https://github.com/pytorch/vision/blob/main/torchvision/models/mnasnet.py", "https://pytorch.org/vision/stable/models/mnasnet.html", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://www.kaggle.com/models/tensorflow/mnasnet", "https://github.com/onnx/models/tree/main/Computer_Vision/mnasnet0_5_Opset16_torch_hub", "https://arxiv.org/html/1807.11626" ], "unverified": [] }, { "id": "mixnet", "name": "MixNet", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "MixNet-S, M and L.", "context_length": null, "release_date": "2019-07-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Mobile classifier using mixed depthwise convolution kernel sizes, from the paper \"MixConv: Mixed Depthwise Convolutional Kernels\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1907.09595", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/mixnet-l", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/onnx/models/tree/main/Computer_Vision/mixnet_l_Opset16_timm" ], "unverified": [] }, { "id": "google-nasnet", "name": "NASNet", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "NASNet-A Mobile and Large.", "context_length": null, "release_date": "2017-07-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Image classifier found by neural architecture search, from the paper \"Learning Transferable Architectures for Scalable Image Recognition\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1707.07012", "https://aihub.qualcomm.com/models/nasnet", "https://github.com/huggingface/pytorch-image-models/tree/main", "https://www.kaggle.com/models/google/nasnet", "https://www.kaggle.com/models/tensorflow/nasnet", "https://www.kaggle.com/models/google/pnasnet", "https://www.kaggle.com/models/google/amoebanet" ], "unverified": [] }, { "id": "google-pnasnet", "name": "PNASNet-5", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-12-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Image classifier from the paper \"Progressive Neural Architecture Search\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1712.00559", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/tensorflow/models/blob/master/research/README.md", "https://www.kaggle.com/models/google/pnasnet", "https://github.com/onnx/models/tree/main/Computer_Vision/pnasnet5large_Opset16_timm" ], "unverified": [] }, { "id": "google-amoebanet", "name": "AmoebaNet-A", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Image classifier found by evolutionary architecture search, published by Google on Kaggle Models.", "verified_on": "2026-10-12", "sources": [ "https://www.kaggle.com/models/google/amoebanet" ], "unverified": [] }, { "id": "google-efficientnet", "name": "EfficientNet", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "EfficientNet B0 to B7, Noisy Student and AdvProp weights, EfficientNet-Lite 0 to 4, EfficientNet-EdgeTPU, EfficientNetV2 S, M, L.", "context_length": null, "release_date": "2019-05-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "mobile or NPU", "CPU server", "microcontroller" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, OpenVINO, Edge Impulse, STM32 model zoo, TI edge AI model zoo, MediaPipe.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google family of image classifiers that scale depth, width and resolution together; one of the most used backbones for crop, defect and medical image classifiers and for EfficientDet.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1905.11946", "https://cloud.google.com/model-garden", "https://huggingface.co/google/efficientnet-b7", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://huggingface.co/timm/tf_efficientnetv2_s.in21k_ft_in1k", "https://aihub.qualcomm.com/models/efficientnet_b0", "https://github.com/pytorch/vision/blob/main/torchvision/models/efficientnet.py" ], "unverified": [] }, { "id": "google-efficientnet-b0", "name": "EfficientNet-B0", "level": "variant", "family_id": "google-efficientnet", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": 5300000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-05-28", "release_date_basis": "announcement page", "weights_date": "2022-12-12", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "The smallest EfficientNet baseline found by architecture search; about 5.3M parameters.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1905.11946", "https://huggingface.co/timm/efficientnet_b0.ra_in1k" ], "unverified": [] }, { "id": "google-efficientnet-b7", "name": "EfficientNet-B7", "level": "variant", "family_id": "google-efficientnet", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": 66000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-05-28", "release_date_basis": "announcement page", "weights_date": "2023-02-15", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "The largest EfficientNet in the original paper; about 66M parameters.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1905.11946", "https://huggingface.co/google/efficientnet-b7" ], "unverified": [] }, { "id": "google-efficientnet-lite", "name": "EfficientNet-Lite", "level": "variant", "family_id": "google-efficientnet", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Lite0 to Lite4.", "context_length": null, "release_date": "2019-05-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, TI edge AI model zoo, MediaPipe.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "EfficientNet variants without squeeze-and-excitation and swish, for mobile CPUs, GPUs and edge accelerators.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1905.11946", "https://aihub.qualcomm.com/models/efficientnet_lite4", "https://github.com/RangiLyu/EfficientNet-Lite", "https://github.com/TexasInstruments/edgeai-tensorlab/tree/main/edgeai-modelzoo/models/vision/classification/imagenet1k", "https://ai.google.dev/edge/mediapipe/solutions/vision/image_classifier", "https://github.com/onnx/models/tree/main/validated/vision/classification/efficientnet-lite4" ], "unverified": [] }, { "id": "google-efficientnetv2", "name": "EfficientNetV2", "level": "variant", "family_id": "google-efficientnet", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": 21612360, "params_active": null, "params_checkpoint": null, "params_note": "B0 to B3, S, M, L, XL.", "context_length": null, "release_date": "2021-04-01", "release_date_basis": "announcement page", "weights_date": "2022-12-13", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Second EfficientNet generation with faster training, from the paper \"EfficientNetV2: Smaller Models and Faster Training\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2104.00298", "https://huggingface.co/timm/tf_efficientnetv2_s.in21k_ft_in1k", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://aihub.qualcomm.com/models/efficientnet_v2_s", "https://github.com/pytorch/vision/blob/main/torchvision/models/efficientnet.py", "https://pytorch.org/vision/stable/models/efficientnetv2.html", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md" ], "unverified": [] }, { "id": "shufflenet", "name": "ShuffleNet", "level": "family", "family_id": null, "developer": [ "megvii" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "ShuffleNet v1 and v2 (x0.5 to x2.0).", "context_length": null, "release_date": "2017-07-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "mobile or NPU", "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Family of efficient mobile CNNs that use channel shuffle and group convolutions.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1707.01083", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://aihub.qualcomm.com/models/shufflenet_v2", "https://github.com/pytorch/vision/blob/main/torchvision/models/shufflenetv2.py", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/shufflenet-v2-x1.0", "https://pytorch.org/hub/pytorch_vision_shufflenet_v2", "https://raw.githubusercontent.com/pytorch/hub/master/pytorch_vision_shufflenet_v2.md", "https://arxiv.org/html/1707.01083" ], "unverified": [] }, { "id": "shufflenet-v2", "name": "ShuffleNet v2", "level": "variant", "family_id": "shufflenet", "developer": [ "megvii" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-07-30", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Second ShuffleNet, designed from practical speed guidelines; from the paper \"ShuffleNet V2\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1807.11164", "https://arxiv.org/html/1807.11164" ], "unverified": [] }, { "id": "squeezenet", "name": "SqueezeNet", "level": "model", "family_id": null, "developer": [ "uc-berkeley" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "SqueezeNet 1.0 and 1.1.", "context_length": null, "release_date": "2016-02-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small image classifier with AlexNet-level accuracy and 50 times fewer parameters (DeepScale and UC Berkeley).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1602.07360", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://aihub.qualcomm.com/models/squeezenet1_1", "https://github.com/pytorch/vision/blob/main/torchvision/models/squeezenet.py", "https://pytorch.org/hub/pytorch_vision_squeezenet", "https://raw.githubusercontent.com/pytorch/hub/master/pytorch_vision_squeezenet.md", "https://modelzoo.co/model/squeezenet-caffe", "https://arxiv.org/html/1602.07360" ], "unverified": [] }, { "id": "huawei-ghostnet", "name": "GhostNet", "level": "model", "family_id": null, "developer": [ "huawei" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "GhostNet v1, v2, v3 and RepGhostNet.", "context_length": null, "release_date": "2019-11-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Efficient mobile CNN from Huawei Noah's Ark Lab that generates extra feature maps with cheap operations.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1911.11907", "https://github.com/huawei-noah/ghostnet", "https://pytorch.org/hub/pytorch_vision_ghostnet", "https://raw.githubusercontent.com/pytorch/hub/master/pytorch_vision_ghostnet.md", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/onnx/models/tree/main/Computer_Vision/ghostnet_100_Opset16_timm" ], "unverified": [] }, { "id": "mit-han-lab-proxylessnas", "name": "ProxylessNAS", "level": "model", "family_id": null, "developer": [ "mit-han-lab" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-12-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Hardware-specialised mobile CNNs found by direct neural architecture search on the target task and device.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1812.00332", "https://github.com/mit-han-lab/ProxylessNAS", "https://pytorch.org/hub/pytorch_vision_proxylessnas", "https://raw.githubusercontent.com/pytorch/hub/master/pytorch_vision_proxylessnas.md" ], "unverified": [] }, { "id": "mit-han-lab-once-for-all", "name": "Once-for-All (OFA)", "level": "model", "family_id": null, "developer": [ "mit-han-lab" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-08-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "One trained supernetwork from which specialised subnetworks are extracted for different devices without retraining.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1908.09791", "https://github.com/mit-han-lab/once-for-all", "https://pytorch.org/hub/pytorch_vision_once_for_all", "https://raw.githubusercontent.com/pytorch/hub/master/pytorch_vision_once_for_all.md" ], "unverified": [] }, { "id": "single-path-nas", "name": "Single-Path NAS", "level": "model", "family_id": null, "developer": [ "carnegie-mellon-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-04-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Hardware-efficient mobile CNN from the paper \"Single-Path NAS: Designing Hardware-Efficient ConvNets in less than 4 Hours\"; timm also ships TinyNet.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1904.02877", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/onnx/models/tree/main/Computer_Vision/spnasnet_100_Opset16_timm", "https://arxiv.org/html/1904.02877" ], "unverified": [] }, { "id": "hardcore-nas", "name": "HardCoRe-NAS", "level": "model", "family_id": null, "developer": [ "alibaba-group" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-02-23", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Latency-constrained mobile classifiers found by differentiable architecture search; timm groups them with FBNet.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2102.11646", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/onnx/models/tree/main/Computer_Vision/fbnetc_100_Opset16_timm", "https://arxiv.org/html/2102.11646" ], "unverified": [] }, { "id": "gpu-efficient-networks", "name": "GPU-Efficient Networks (GENet)", "level": "model", "family_id": null, "developer": [ "alibaba-group" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-06-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Image classifiers designed for high GPU throughput, from the paper \"Neural Architecture Design for GPU-Efficient Networks\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2006.14090", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/onnx/models/tree/main/Computer_Vision/gernet_l_Opset16_timm", "https://arxiv.org/html/2006.14090" ], "unverified": [] }, { "id": "nvidia-gpunet", "name": "GPUNet", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA image classifiers optimised for inference latency on NVIDIA GPUs, published in Deep Learning Examples.", "verified_on": "2026-10-12", "sources": [ "https://github.com/NVIDIA/DeepLearningExamples/tree/master/PyTorch/Classification/GPUNet", "https://github.com/NVIDIA/DeepLearningExamples", "https://aihub.qualcomm.com/models/gpunet", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/dle/models/gpunet_0_pyt_ckpt", "https://pytorch.org/hub/nvidia_deeplearningexamples_gpunet", "https://raw.githubusercontent.com/pytorch/hub/master/nvidia_deeplearningexamples_gpunet.md", "https://arxiv.org/abs/2205.00841" ], "unverified": [] }, { "id": "rexnet", "name": "ReXNet", "level": "model", "family_id": null, "developer": [ "naver" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-07-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Mobile image classifier with redesigned channel dimensions, from the paper \"Rethinking Channel Dimensions for Efficient Model Design\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2007.00992", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/rexnet-v1-x1.0", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/onnx/models/tree/main/Computer_Vision/rexnet_100_Opset16_timm", "https://arxiv.org/html/2007.00992" ], "unverified": [] }, { "id": "baidu-pp-lcnet", "name": "PP-LCNet", "level": "model", "family_id": null, "developer": [ "baidu" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "PP-LCNet and PP-LCNetV2; PaddleOCR uses PP-LCNet for document and text-line orientation.", "context_length": null, "release_date": "2021-09-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Lightweight CPU-oriented CNN from Baidu PaddlePaddle, from the paper \"PP-LCNet: A Lightweight CPU Convolutional Neural Network\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2109.15099", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/onnx/models/tree/main/Computer_Vision/lcnet_075_Opset16_timm" ], "unverified": [] }, { "id": "hgnet", "name": "PP-HGNet", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "HGNet and HGNetV2 (backbone of RT-DETR).", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "High-performance GPU CNN backbone from Baidu PaddlePaddle; HGNetV2 is published in KerasHub on Kaggle.", "verified_on": "2026-10-12", "sources": [ "https://www.kaggle.com/models/keras/hgnetv2", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://keras.io/keras_hub/presets/" ], "unverified": [] }, { "id": "selecsls", "name": "SelecSLS", "level": "model", "family_id": null, "developer": [ "max-planck-institute-for-informatics" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-07-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fast CNN backbone introduced in the real-time 3D motion capture paper \"XNect\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1907.00837", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/onnx/models/tree/main/Computer_Vision/selecsls42b_Opset16_timm", "https://arxiv.org/html/1907.00837" ], "unverified": [] }, { "id": "mobileone", "name": "MobileOne", "level": "model", "family_id": null, "developer": [ "apple" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-06-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Mobile backbone with re-parameterisable branches that runs under one millisecond on an iPhone 12.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2206.04040", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/mobileone", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2206.04040" ], "unverified": [] }, { "id": "fasternet", "name": "FasterNet", "level": "model", "family_id": null, "developer": [ "hkust" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-03-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fast CNN using partial convolution, from the paper \"Run, Don't Walk: Chasing Higher FLOPS for Faster Neural Networks\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2303.03667", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2303.03667" ], "unverified": [] }, { "id": "starnet", "name": "StarNet", "level": "model", "family_id": null, "developer": [ "northeastern-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-03-29", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Efficient network built on element-wise multiplication, from the paper \"Rewrite the Stars\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2403.19967", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2403.19967" ], "unverified": [] }, { "id": "cpubone", "name": "CPUBone", "level": "model", "family_id": null, "developer": [ "university-of-udine" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-03-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vision backbone designed for devices with low parallelisation such as CPUs, from the paper \"CPUBone\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2603.26425", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2603.26425" ], "unverified": [] }, { "id": "convnext-family", "name": "ConvNeXt", "level": "family", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "ConvNeXt T, S, B, L, XL; ConvNeXt V2 from Atto to Huge.", "context_length": null, "release_date": "2022-01-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "mobile or NPU", "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta family of modernised pure convolutional networks that match vision transformers on ImageNet, COCO and ADE20K.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2201.03545", "https://ai.azure.com/catalog/models?task=image-classification", "https://huggingface.co/facebook/convnext-large-224", "https://aihub.qualcomm.com/models/convnext_base", "https://github.com/pytorch/vision/blob/main/torchvision/models/convnext.py", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/convnext-tiny", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/pretrained_convnextv2" ], "unverified": [] }, { "id": "meta-convnext", "name": "ConvNeXt", "level": "variant", "family_id": "convnext-family", "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-01-10", "release_date_basis": "announcement page", "weights_date": "2022-03-02", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First ConvNeXt generation, from the paper \"A ConvNet for the 2020s\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2201.03545", "https://huggingface.co/facebook/convnext-large-224" ], "unverified": [] }, { "id": "meta-convnext-v2", "name": "ConvNeXt V2", "level": "variant", "family_id": "convnext-family", "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": 197956840, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-01-02", "release_date_basis": "announcement page", "weights_date": "2023-01-05", "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ConvNeXt pretrained with a fully convolutional masked autoencoder, from the paper \"ConvNeXt V2\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2301.00808", "https://huggingface.co/timm/convnextv2_large.fcmae_ft_in22k_in1k_384" ], "unverified": [] }, { "id": "replknet", "name": "RepLKNet", "level": "model", "family_id": null, "developer": [ "tsinghua-university", "megvii" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-03-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "CNN with very large (31x31) kernels, from the paper \"Scaling Up Your Kernels to 31x31\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2203.06717", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/replknet", "https://arxiv.org/html/2203.06717" ], "unverified": [] }, { "id": "hornet", "name": "HorNet", "level": "model", "family_id": null, "developer": [ "tsinghua-university", "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-07-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Backbone with recursive gated convolutions for high-order spatial interactions.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2207.14284", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/hornet", "https://arxiv.org/html/2207.14284" ], "unverified": [] }, { "id": "van", "name": "VAN (Visual Attention Network)", "level": "model", "family_id": null, "developer": [ "tsinghua-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-02-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Backbone using large kernel attention, from the paper \"Visual Attention Network\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2202.09741", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/van", "https://arxiv.org/html/2202.09741" ], "unverified": [] }, { "id": "focalnet", "name": "FocalNet", "level": "model", "family_id": null, "developer": [ "microsoft-research" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-03-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Backbone that replaces self-attention with focal modulation, from the paper \"Focal Modulation Networks\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2203.11926", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2203.11926" ], "unverified": [] }, { "id": "inceptionnext", "name": "InceptionNeXt", "level": "model", "family_id": null, "developer": [ "nus", "sea-ai-lab" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-03-29", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ConvNeXt variant with Inception-style depthwise convolutions.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2303.16900", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2303.16900" ], "unverified": [] }, { "id": "mambaout", "name": "MambaOut", "level": "model", "family_id": null, "developer": [ "nus" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-05-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Gated CNN backbone from the paper \"MambaOut: Do We Really Need Mamba for Vision?\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2405.07992", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2405.07992" ], "unverified": [] }, { "id": "metaformer", "name": "MetaFormer (PoolFormer, ConvFormer, CAFormer)", "level": "model", "family_id": null, "developer": [ "sea-ai-lab", "nus" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": 11915176, "params_active": null, "params_checkpoint": null, "params_note": "PoolFormer v1 and v2, ConvFormer and CAFormer (arXiv 2210.13452).", "context_length": null, "release_date": "2021-11-22", "release_date_basis": "announcement page", "weights_date": "2022-03-02", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Sea AI Lab family that keeps the transformer block layout but swaps the token mixer, starting with pooling in PoolFormer.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2111.11418", "https://huggingface.co/sail/poolformer_s12", "https://ai.azure.com/catalog/models?task=image-classification", "https://github.com/open-mmlab/mmsegmentation/tree/main/configs/poolformer", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/abs/2210.13452", "https://arxiv.org/html/2111.11418" ], "unverified": [] }, { "id": "convmixer", "name": "ConvMixer", "level": "model", "family_id": null, "developer": [ "carnegie-mellon-university", "bosch" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-01-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Simple patch-based convolutional classifier from the paper \"Patches Are All You Need?\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2201.09792", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/convmixer", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://www.kaggle.com/models/rishitdagli/convmixer", "https://github.com/onnx/models/tree/main/Computer_Vision/convmixer_1024_20_ks9_p14_Opset16_timm", "https://arxiv.org/html/2201.09792" ], "unverified": [] }, { "id": "sequencer2d", "name": "Sequencer2D", "level": "model", "family_id": null, "developer": [ "rikkyo-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-05-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Image classifier that uses LSTMs instead of self-attention, from the paper \"Sequencer: Deep LSTM for Image Classification\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2205.01972", "https://aihub.qualcomm.com/models/sequencer2d", "https://github.com/okojoalg/sequencer", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/onnx/models/tree/main/Computer_Vision/sequencer2d_l_Opset16_timm", "https://arxiv.org/html/2205.01972" ], "unverified": [] }, { "id": "mlp-mixer", "name": "MLP-Mixer", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "timm also ships gMLP and ResMLP in the same group.", "context_length": null, "release_date": "2021-05-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "All-MLP image classifier, from the paper \"MLP-Mixer: An all-MLP Architecture for Vision\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2105.01601", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/mlp_mixer", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://www.kaggle.com/models/spsayakpaul/mlp-mixer", "https://github.com/onnx/models/tree/main/Computer_Vision/gmixer_24_224_Opset16_timm", "https://arxiv.org/html/2105.01601" ], "unverified": [] }, { "id": "huawei-snn-mlp", "name": "SNN-MLP", "level": "model", "family_id": null, "developer": [ "huawei" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-03-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MLP backbone inspired by spiking neurons, from the paper \"Brain-inspired Multilayer Perceptron with Spiking Neurons\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2203.14679", "https://github.com/huawei-noah/Efficient-AI-Backbones", "https://pytorch.org/hub/pytorch_vision_snnmlp", "https://raw.githubusercontent.com/pytorch/hub/master/pytorch_vision_snnmlp.md" ], "unverified": [] }, { "id": "google-vit", "name": "Vision Transformer (ViT)", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "ViT-Ti, S, B, L and H with 8, 14, 16 and 32 pixel patches; AugReg weights (arXiv 2106.10270).", "context_length": null, "release_date": "2020-10-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "mobile or NPU", "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google transformer image classifier that treats an image as a sequence of 16x16 patches; the base of most later vision transformers.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2010.11929", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://huggingface.co/google/vit-base-patch16-224", "https://aihub.qualcomm.com/models/vit", "https://github.com/pytorch/vision/blob/main/torchvision/models/vision_transformer.py", "https://huggingface.co/OpenVINO/vit_tiny_cls-fp16-ov", "https://huggingface.co/api/models?pipeline_tag=image-classification&sort=downloads&limit=50" ], "unverified": [] }, { "id": "google-vit-base-patch16-224", "name": "ViT-B/16", "level": "variant", "family_id": "google-vit", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 86000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-10-22", "release_date_basis": "announcement page", "weights_date": "2022-03-02", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Base ViT with 16x16 patches, pretrained on ImageNet-21k and fine-tuned on ImageNet at 224 pixels.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2010.11929", "https://huggingface.co/google/vit-base-patch16-224" ], "unverified": [] }, { "id": "google-vit-large-patch16-224", "name": "ViT-L/16", "level": "variant", "family_id": "google-vit", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 304000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-10-22", "release_date_basis": "announcement page", "weights_date": "2022-03-02", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Large ViT with 16x16 patches pretrained on ImageNet-21k.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2010.11929", "https://huggingface.co/google/vit-large-patch16-224-in21k" ], "unverified": [] }, { "id": "google-vit-augreg", "name": "ViT AugReg", "level": "variant", "family_id": "google-vit", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 86567656, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-06-18", "release_date_basis": "announcement page", "weights_date": "2022-12-22", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ViT checkpoints trained with the data augmentation and regularisation recipe of \"How to train your ViT?\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2106.10270", "https://huggingface.co/timm/vit_base_patch16_224.augreg2_in21k_ft_in1k" ], "unverified": [] }, { "id": "meta-deit", "name": "DeiT", "level": "family", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "DeiT Ti, S, B (with distillation token) and DeiT III.", "context_length": null, "release_date": "2020-12-23", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: TI edge AI model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta data-efficient image transformers trained on ImageNet alone with distillation through attention.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2012.12877", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://huggingface.co/facebook/deit-base-distilled-patch16-224", "https://github.com/TexasInstruments/edgeai-tensorlab/tree/main/edgeai-modelzoo/models/vision/classification/imagenet1k", "https://huggingface.co/api/models?pipeline_tag=image-classification&sort=downloads&limit=50", "https://huggingface.co/timm/deit_tiny_patch16_224.fb_in1k", "https://www.kaggle.com/models/keras/deit" ], "unverified": [] }, { "id": "meta-deit-base-distilled", "name": "DeiT-B distilled", "level": "variant", "family_id": "meta-deit", "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-12-23", "release_date_basis": "announcement page", "weights_date": "2022-03-02", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Base DeiT with a distillation token learned from a CNN teacher.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2012.12877", "https://huggingface.co/facebook/deit-base-distilled-patch16-224" ], "unverified": [] }, { "id": "meta-deit-tiny", "name": "DeiT-Ti", "level": "variant", "family_id": "meta-deit", "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 5717416, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-12-23", "release_date_basis": "announcement page", "weights_date": "2023-03-28", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "The smallest DeiT, about 5M parameters.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2012.12877", "https://huggingface.co/timm/deit_tiny_patch16_224.fb_in1k" ], "unverified": [] }, { "id": "meta-deit-iii", "name": "DeiT III", "level": "variant", "family_id": "meta-deit", "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-04-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Improved supervised ViT training recipe, from the paper \"DeiT III: Revenge of the ViT\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2204.07118" ], "unverified": [] }, { "id": "cait", "name": "CaiT", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 356460520, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-03-31", "release_date_basis": "announcement page", "weights_date": "2023-04-13", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Deeper image transformers with class-attention layers, from the paper \"Going deeper with Image Transformers\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.17239", "https://huggingface.co/timm/cait_m48_448.fb_dist_in1k", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://www.kaggle.com/models/spsayakpaul/cait", "https://github.com/onnx/models/tree/main/Computer_Vision/cait_m48_448_Opset16_timm", "https://arxiv.org/html/2103.17239" ], "unverified": [] }, { "id": "meta-levit", "name": "LeViT", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "hybrid", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-04-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU", "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Hybrid convolution and transformer classifier designed for fast inference, from the paper \"LeViT: a Vision Transformer in ConvNet's Clothing\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2104.01136", "https://github.com/facebookresearch/LeViT", "https://aihub.qualcomm.com/models/levit", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/levit-128s", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/levit", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/onnx/models/tree/main/Computer_Vision/levit_128s_Opset16_timm" ], "unverified": [] }, { "id": "convit", "name": "ConViT", "level": "model", "family_id": null, "developer": [ "ens-paris", "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-03-19", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vision transformer with soft convolutional inductive biases.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.10697", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/onnx/models/tree/main/Computer_Vision/convit_small_Opset16_timm", "https://arxiv.org/html/2103.10697" ], "unverified": [] }, { "id": "xcit", "name": "XCiT", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-06-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Cross-covariance image transformers whose attention scales linearly with the number of tokens.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2106.09681", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/xcit", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://www.kaggle.com/models/rishitdagli/xcit", "https://github.com/onnx/models/tree/main/Computer_Vision/xcit_large_24_p16_224_Opset16_timm", "https://arxiv.org/html/2106.09681" ], "unverified": [] }, { "id": "microsoft-swin-transformer", "name": "Swin Transformer", "level": "family", "family_id": null, "developer": [ "microsoft" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Swin T, S, B, L; Swin V2 up to 3B parameters; Video Swin is a separate record.", "context_length": null, "release_date": "2021-03-25", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "mobile or NPU", "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft hierarchical vision transformer with shifted windows; a general backbone for classification, detection and segmentation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.14030", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://huggingface.co/microsoft/swin-base-patch4-window7-224", "https://ai.azure.com/catalog/models?task=image-classification", "https://huggingface.co/microsoft/swinv2-tiny-patch4-window16-256", "https://aihub.qualcomm.com/models/swin_base", "https://github.com/pytorch/vision/blob/main/torchvision/models/swin_transformer.py" ], "unverified": [] }, { "id": "microsoft-swin-base", "name": "Swin-B", "level": "variant", "family_id": "microsoft-swin-transformer", "developer": [ "microsoft" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 87825848, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-03-25", "release_date_basis": "announcement page", "weights_date": "2022-03-02", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Base Swin Transformer with 7x7 windows at 224 pixels.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.14030", "https://huggingface.co/microsoft/swin-base-patch4-window7-224" ], "unverified": [] }, { "id": "microsoft-swin-tiny", "name": "Swin-T", "level": "variant", "family_id": "microsoft-swin-transformer", "developer": [ "microsoft" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-03-25", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "The smallest Swin Transformer, also shipped in the OpenVINO Open Model Zoo.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.14030", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/swin-tiny-patch4-window7-224" ], "unverified": [] }, { "id": "microsoft-swin-v2", "name": "Swin Transformer V2", "level": "variant", "family_id": "microsoft-swin-transformer", "developer": [ "microsoft" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-11-18", "release_date_basis": "announcement page", "weights_date": "2022-06-14", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Swin scaled to higher capacity and resolution, from the paper \"Swin Transformer V2: Scaling Up Capacity and Resolution\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2111.09883", "https://huggingface.co/microsoft/swinv2-tiny-patch4-window16-256" ], "unverified": [] }, { "id": "swin-s3", "name": "Swin S3 (AutoFormerV2)", "level": "model", "family_id": null, "developer": [ "stony-brook-university", "microsoft-research" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-11-29", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vision transformers found by searching the search space, from the paper \"Searching the Search Space of Vision Transformer\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2111.14725", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2111.14725" ], "unverified": [] }, { "id": "microsoft-beit", "name": "BEiT", "level": "family", "family_id": null, "developer": [ "microsoft" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "BEiT, BEiT v2 and BEiT-3.", "context_length": null, "release_date": "2021-06-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft vision transformers pretrained by masked image modelling in the style of BERT.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2106.08254", "https://ai.azure.com/catalog/models?task=image-classification", "https://huggingface.co/microsoft/beit-base-patch16-224-pt22k-ft22k", "https://aihub.qualcomm.com/models/beit", "https://github.com/microsoft/unilm/tree/master/beit", "https://huggingface.co/api/models?pipeline_tag=image-classification&sort=likes&limit=50", "https://www.modelscope.cn/models/iic/cv_beitv2-large_image-classification_patch16_224_pt1k_ft22k_in1k" ], "unverified": [] }, { "id": "microsoft-beit-base", "name": "BEiT-B", "level": "variant", "family_id": "microsoft-beit", "developer": [ "microsoft" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-06-15", "release_date_basis": "announcement page", "weights_date": "2022-03-02", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Base BEiT pretrained and fine-tuned on ImageNet-22k.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2106.08254", "https://huggingface.co/microsoft/beit-base-patch16-224-pt22k-ft22k" ], "unverified": [] }, { "id": "microsoft-beit-v2", "name": "BEiT v2", "level": "variant", "family_id": "microsoft-beit", "developer": [ "microsoft" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-08-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "BEiT with vector-quantised visual tokenisers, from the paper \"BEiT v2\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2208.06366" ], "unverified": [] }, { "id": "microsoft-beit-3", "name": "BEiT-3", "level": "variant", "family_id": "microsoft-beit", "developer": [ "microsoft" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-08-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "General vision and vision-language foundation model, from the paper \"Image as a Foreign Language\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2208.10442" ], "unverified": [] }, { "id": "microsoft-cvt", "name": "CvT", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "hybrid", "architecture_detail": null, "params_total": 20023208, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-03-29", "release_date_basis": "announcement page", "weights_date": "2022-04-04", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vision transformer with convolutional token embedding and projection, from the paper \"CvT: Introducing Convolutions to Vision Transformers\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.15808", "https://huggingface.co/microsoft/cvt-13", "https://ai.azure.com/catalog/models?task=image-classification" ], "unverified": [] }, { "id": "pvt", "name": "PVT (Pyramid Vision Transformer)", "level": "model", "family_id": null, "developer": [ "nanjing-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "PVT and PVT v2 (arXiv 2106.13797).", "context_length": null, "release_date": "2021-02-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Pyramid vision transformer backbone for dense prediction without convolutions.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2102.12122", "https://github.com/open-mmlab/mmdetection/tree/main/configs/pvt", "https://arxiv.org/abs/2106.13797", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2102.12122" ], "unverified": [] }, { "id": "twins", "name": "Twins", "level": "model", "family_id": null, "developer": [ "meituan", "university-of-adelaide" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Twins-PCPVT and Twins-SVT.", "context_length": null, "release_date": "2021-04-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vision transformers with redesigned spatial attention, from the paper \"Twins\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2104.13840", "https://github.com/open-mmlab/mmsegmentation/tree/main/configs/twins", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/onnx/models/tree/main/Computer_Vision/twins_pcpvt_base_Opset17_timm", "https://arxiv.org/html/2104.13840" ], "unverified": [] }, { "id": "t2t-vit", "name": "T2T-ViT", "level": "model", "family_id": null, "developer": [ "nus" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-01-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vision transformer trained from scratch on ImageNet with tokens-to-token embedding.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2101.11986", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/t2t-vit-14", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/t2t_vit", "https://arxiv.org/html/2101.11986" ], "unverified": [] }, { "id": "tnt", "name": "TNT (Transformer in Transformer)", "level": "model", "family_id": null, "developer": [ "huawei-technologies" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-02-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vision transformer that models both patch-level and pixel-level features.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.00112", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/tnt", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://www.kaggle.com/models/rishitdagli/tnt-s", "https://github.com/onnx/models/tree/main/Computer_Vision/tnt_s_patch16_224_Opset16_timm", "https://arxiv.org/html/2103.00112" ], "unverified": [] }, { "id": "crossvit", "name": "CrossViT", "level": "model", "family_id": null, "developer": [ "ibm" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-03-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Dual-branch multi-scale vision transformer with cross-attention.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.14899", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/onnx/models/tree/main/Computer_Vision/crossvit_15_dagger_408_Opset16_timm", "https://arxiv.org/html/2103.14899" ], "unverified": [] }, { "id": "pit", "name": "PiT (Pooling-based ViT)", "level": "model", "family_id": null, "developer": [ "naver" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-03-30", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vision transformer with spatial pooling stages, from the paper \"Rethinking Spatial Dimensions of Vision Transformers\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.16302", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/onnx/models/tree/main/Computer_Vision/pit_b_224_Opset16_timm", "https://arxiv.org/html/2103.16302" ], "unverified": [] }, { "id": "coat", "name": "CoaT", "level": "model", "family_id": null, "developer": [ "uc-san-diego" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-04-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Co-scale conv-attentional image transformers.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2104.06399", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/abs/2106.04803", "https://github.com/onnx/models/tree/main/Computer_Vision/coat_lite_small_Opset16_timm", "https://arxiv.org/html/2104.06399" ], "unverified": [] }, { "id": "visformer", "name": "Visformer", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "hybrid", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-04-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vision-friendly transformer combining convolution and attention.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2104.12533", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/onnx/models/tree/main/Computer_Vision/visformer_small_Opset16_timm" ], "unverified": [] }, { "id": "nest", "name": "NesT", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-05-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Nested hierarchical transformer, from the paper \"Nested Hierarchical Transformer\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2105.12723", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/onnx/models/tree/main/Computer_Vision/jx_nest_base_Opset16_timm" ], "unverified": [] }, { "id": "coatnet", "name": "CoAtNet", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "hybrid", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-06-09", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Hybrid of convolution and attention, from the paper \"CoAtNet: Marrying Convolution and Attention for All Data Sizes\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2106.04803", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2106.04803" ], "unverified": [] }, { "id": "volo", "name": "VOLO", "level": "model", "family_id": null, "developer": [ "sea-ai-lab", "nus" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-06-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vision outlooker transformer, from the paper \"VOLO: Vision Outlooker for Visual Recognition\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2106.13112", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2106.13112" ], "unverified": [] }, { "id": "maxvit", "name": "MaxViT", "level": "model", "family_id": null, "developer": [ "google", "ut-austin" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "hybrid", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-04-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multi-axis vision transformer with blocked local and dilated global attention.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2204.01697", "https://pytorch.org/vision/stable/models/maxvit.html", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/tensorflow/models/tree/master/official/projects/maxvit", "https://arxiv.org/html/2204.01697" ], "unverified": [] }, { "id": "davit", "name": "DaViT", "level": "model", "family_id": null, "developer": [ "hku", "microsoft-corporation" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-04-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Dual attention vision transformer with spatial and channel attention.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2204.03645", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/davit", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2204.03645" ], "unverified": [] }, { "id": "gcvit", "name": "GCViT", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "hybrid", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-06-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Global context vision transformer, from the paper \"Global Context Vision Transformers\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2206.09959", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2206.09959" ], "unverified": [] }, { "id": "apple-mobilevit", "name": "MobileViT", "level": "model", "family_id": null, "developer": [ "apple" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "hybrid", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "MobileViT XXS, XS, S and MobileViTv2; DeepLabV3 segmentation heads.", "context_length": null, "release_date": "2021-10-05", "release_date_basis": "announcement page", "weights_date": "2022-05-30", "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Apple light-weight hybrid of convolution and transformer for mobile classification and segmentation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2110.02178", "https://huggingface.co/apple/mobilevit-small", "https://ai.azure.com/catalog/models?task=image-classification", "https://huggingface.co/apple/mobilevit-xx-small", "https://aihub.qualcomm.com/models/mobile_vit", "https://github.com/apple/ml-cvnets", "https://huggingface.co/api/models?pipeline_tag=image-classification&sort=likes&limit=50" ], "unverified": [] }, { "id": "fastvit", "name": "FastViT", "level": "model", "family_id": null, "developer": [ "apple" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "hybrid", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-03-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Apple Core ML.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fast hybrid vision transformer using structural re-parameterisation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2303.14189", "https://developer.apple.com/machine-learning/models/", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2303.14189" ], "unverified": [] }, { "id": "snap-efficientformer", "name": "EfficientFormer", "level": "model", "family_id": null, "developer": [ "snap", "northeastern-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "EfficientFormer v1 and v2.", "context_length": null, "release_date": "2022-06-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vision transformers that run at MobileNet speed on phones, from Snap Research.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2206.01191", "https://github.com/snap-research/EfficientFormer", "https://aihub.qualcomm.com/models/efficientformer", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/efficientformer", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2206.01191" ], "unverified": [] }, { "id": "efficientvit-mla", "name": "EfficientViT (multi-scale linear attention)", "level": "model", "family_id": null, "developer": [ "massachusetts-institute-of-technology" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-05-29", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MIT Han Lab vision transformer with multi-scale linear attention for high-resolution dense prediction.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2205.14756", "https://aihub.qualcomm.com/models/efficientvit_b2_cls", "https://github.com/CVHub520/efficientvit", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2205.14756" ], "unverified": [] }, { "id": "efficientvit-cga", "name": "EfficientViT (cascaded group attention)", "level": "model", "family_id": null, "developer": [ "cuhk", "microsoft-research" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-05-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Memory-efficient vision transformer with cascaded group attention.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2305.07027", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2305.07027" ], "unverified": [] }, { "id": "edgenext", "name": "EdgeNeXt", "level": "model", "family_id": null, "developer": [ "mbzuai" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "hybrid", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-06-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Hybrid CNN and transformer for mobile vision applications.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2206.10589", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/edgenext", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/onnx/models/tree/main/Computer_Vision/edgenext_small_Opset16_timm", "https://arxiv.org/html/2206.10589" ], "unverified": [] }, { "id": "next-vit", "name": "Next-ViT", "level": "model", "family_id": null, "developer": [ "bytedance" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "hybrid", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-07-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Hybrid vision transformer designed for deployment in industrial scenarios, from the paper \"Next-ViT\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2207.05501", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2207.05501" ], "unverified": [] }, { "id": "tinyvit", "name": "TinyViT", "level": "model", "family_id": null, "developer": [ "microsoft-research" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-07-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small vision transformers trained by fast pretraining distillation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2207.10666", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2207.10666" ], "unverified": [] }, { "id": "swiftformer", "name": "SwiftFormer", "level": "model", "family_id": null, "developer": [ "mbzuai" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-03-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Mobile vision transformer with efficient additive attention.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2303.15446", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2303.15446" ], "unverified": [] }, { "id": "tsinghua-university-repvit", "name": "RepViT", "level": "model", "family_id": null, "developer": [ "tsinghua-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-07-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Mobile CNN redesigned from a vision transformer perspective, from Tsinghua University.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2307.09283", "https://github.com/THU-MIG/RepViT", "https://aihub.qualcomm.com/models/repvit", "https://www.modelscope.cn/models/THU-MIG/repvit_m1_1.dist_450e_in1k", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md" ], "unverified": [] }, { "id": "shvit", "name": "SHViT", "level": "model", "family_id": null, "developer": [ "university-of-seoul" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-01-29", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Single-head vision transformer with a memory-efficient macro design.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2401.16456", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2401.16456" ], "unverified": [] }, { "id": "hiera", "name": "Hiera", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-06-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Hierarchical vision transformer without extra components, pretrained with MAE; the image encoder of SAM 2.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2306.00989", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md" ], "unverified": [] }, { "id": "rope-vit", "name": "RoPE-ViT", "level": "model", "family_id": null, "developer": [ "naver" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-03-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vision transformer with rotary position embedding, from the paper \"Rotary Position Embedding for Vision Transformer\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2403.13298", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2403.13298" ], "unverified": [] }, { "id": "vitamin", "name": "ViTamin", "level": "model", "family_id": null, "developer": [ "jhu" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "hybrid", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-04-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vision backbone designed for the vision-language era, from the paper \"ViTamin\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2404.02132", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2404.02132" ], "unverified": [] }, { "id": "navit", "name": "NaViT and FlexiViT (NaFlexViT)", "level": "model", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-07-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vision transformer that handles any aspect ratio and resolution by packing patches, from the paper \"Patch n' Pack: NaViT\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2307.06304", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2307.06304" ], "unverified": [] }, { "id": "conformer-vision", "name": "Conformer (vision)", "level": "model", "family_id": null, "developer": [ "university-of-chinese-academy-of-sciences", "peng-cheng-laboratory" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "hybrid", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-05-09", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Image classifier coupling convolution local features with transformer global representations.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2105.03889", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/conformer", "https://arxiv.org/html/2105.03889" ], "unverified": [] }, { "id": "bottleneck-transformers", "name": "Bottleneck Transformers (BoTNet)", "level": "model", "family_id": null, "developer": [ "uc-berkeley", "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "hybrid", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "timm groups BoTNet with HaloNet and LambdaNet.", "context_length": null, "release_date": "2021-01-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ResNet with self-attention in the bottleneck blocks, from the paper \"Bottleneck Transformers for Visual Recognition\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2101.11605", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/onnx/models/tree/main/Computer_Vision/botnet26t_256_Opset16_timm", "https://arxiv.org/html/2101.11605" ], "unverified": [] }, { "id": "riformer", "name": "RIFormer", "level": "model", "family_id": null, "developer": [ "shanghai-ai-lab", "hku" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-04-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vision backbone that removes the token mixer at inference, from the paper \"RIFormer\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2304.05659", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/riformer", "https://arxiv.org/html/2304.05659" ], "unverified": [] }, { "id": "internimage", "name": "InternImage", "level": "model", "family_id": null, "developer": [ "shanghai-ai-lab", "tsinghua-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Large CNN foundation backbone with deformable convolution; Qualcomm AI Hub ships a phone build.", "verified_on": "2026-10-12", "sources": [ "https://aihub.qualcomm.com/models/internimage", "https://arxiv.org/html/2211.05778" ], "unverified": [] }, { "id": "internvit", "name": "InternViT", "level": "model", "family_id": null, "developer": [ "shanghai-ai-lab", "nanjing-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-12-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vision encoder of the InternVL models, scaled to 6B parameters.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2312.14238", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2312.14238" ], "unverified": [] }, { "id": "eva", "name": "EVA", "level": "family", "family_id": null, "developer": [ "baai", "hust" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "EVA (1B), EVA-02 (Ti to E) and EVA-CLIP.", "context_length": null, "release_date": "2022-11-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vision foundation models pretrained by masked image modelling against CLIP features.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2211.07636", "https://huggingface.co/api/models?pipeline_tag=image-classification&sort=downloads&limit=50", "https://huggingface.co/timm/eva02_base_patch14_448.mim_in22k_ft_in22k_in1k", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/eva02", "https://arxiv.org/abs/2303.11331", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/eva", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2211.07636" ], "unverified": [] }, { "id": "eva-02", "name": "EVA-02", "level": "variant", "family_id": "eva", "developer": [ "baai", "hust" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": 87117544, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-03-20", "release_date_basis": "announcement page", "weights_date": "2023-03-31", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Second EVA generation, from the paper \"EVA-02: A Visual Representation for Neon Genesis\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2303.11331", "https://huggingface.co/timm/eva02_base_patch14_448.mim_in22k_ft_in22k_in1k", "https://arxiv.org/html/2303.11331" ], "unverified": [] }, { "id": "apple-aim", "name": "AIM", "level": "family", "family_id": null, "developer": [ "apple" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "AIM 600M to 7B; AIMv2 large, huge, 1B and 3B.", "context_length": null, "release_date": "2024-01-16", "release_date_basis": "announcement page", "weights_date": "2024-01-11", "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Apple vision encoders pretrained autoregressively, from the paper \"Scalable Pre-training of Large Autoregressive Image Models\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2401.08541", "https://huggingface.co/apple/AIM", "https://huggingface.co/apple/aimv2-large-patch14-224", "https://huggingface.co/apple/AIM-600M", "https://huggingface.co/api/models?pipeline_tag=image-classification&sort=likes&limit=50", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/abs/2411.14402" ], "unverified": [] }, { "id": "apple-aimv2", "name": "AIMv2", "level": "variant", "family_id": "apple-aim", "developer": [ "apple" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": 309197824, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-11-21", "release_date_basis": "announcement page", "weights_date": "2024-10-29", "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Apple vision encoders pretrained with multimodal autoregressive objectives, from the paper \"Multimodal Autoregressive Pre-training of Large Vision Encoders\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2411.14402", "https://huggingface.co/apple/aimv2-large-patch14-224" ], "unverified": [] }, { "id": "eupe", "name": "EUPE (Efficient Universal Perception Encoder)", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-03-23", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Efficient vision encoder for many perception tasks, from the paper \"Efficient Universal Perception Encoder\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2603.22387", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2603.22387" ], "unverified": [] }, { "id": "lingbot-vision", "name": "LingBot-Vision", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-07-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vision backbone pretrained for dense spatial perception, from the paper \"Vision Pretraining for Dense Spatial Perception\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2607.05247", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md" ], "unverified": [] }, { "id": "google-tipsv2", "name": "TIPSv2", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image", "text" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": 195948288, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04-13", "release_date_basis": "announcement page", "weights_date": "2026-04-09", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google image-text encoder with improved patch-text alignment, for zero-shot classification and dense tasks such as depth.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2604.12012", "https://huggingface.co/google/tipsv2-b14", "https://huggingface.co/api/models?pipeline_tag=zero-shot-image-classification&sort=likes&limit=50" ], "unverified": [] }, { "id": "align", "name": "ALIGN", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image", "text" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-02-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Image-text dual encoder trained on noisy alt-text pairs, from the paper \"Scaling Up Visual and Vision-Language Representation Learning With Noisy Text Supervision\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2102.05918", "https://paperswithcode.com/method/align", "https://huggingface.co/datasets/pwc-archive/methods", "https://arxiv.org/abs/2102.05918v2", "https://arxiv.org/html/2102.05918" ], "unverified": [] }, { "id": "microsoft-llm2clip", "name": "LLM2CLIP", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "vision", "modalities_in": [ "image", "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 7504924672, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-11-07", "release_date_basis": "announcement page", "weights_date": "2024-11-16", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft method and checkpoints that use a large language model as the CLIP text encoder for richer image-text representations.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2411.04997", "https://huggingface.co/microsoft/LLM2CLIP-Llama-3-8B-Instruct-CC-Finetuned", "https://huggingface.co/api/models?pipeline_tag=zero-shot-classification&sort=downloads&limit=50" ], "unverified": [] }, { "id": "fashion-clip", "name": "FashionCLIP", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image", "text" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 151277439, "params_active": null, "params_checkpoint": null, "params_note": "Marqo FashionSigLIP (Marqo/marqo-fashionSigLIP) is a related fashion encoder.", "context_length": null, "release_date": "2023-02-21", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "CLIP fine-tuned on fashion product images and text, for product search and tagging in retail.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/patrickjohncyh/fashion-clip", "https://huggingface.co/api/models?pipeline_tag=zero-shot-image-classification&sort=downloads&limit=50" ], "unverified": [] }, { "id": "streetclip", "name": "StreetCLIP", "level": "model", "family_id": null, "developer": [ "stanford-university" ], "model_class": "geospatial", "modalities_in": [ "image", "text" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-02-01", "release_date_basis": "announcement page", "weights_date": "2023-01-26", "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "CLIP model adapted for zero-shot image geolocalisation of street-level photos, from the paper \"Learning Generalized Zero-Shot Learners for Open-Domain Image Geolocalization\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2302.00275", "https://huggingface.co/geolocal/StreetCLIP", "https://huggingface.co/api/models?pipeline_tag=zero-shot-image-classification&sort=likes&limit=50", "https://arxiv.org/html/2302.00275" ], "unverified": [] }, { "id": "meta-dino", "name": "DINO (v1)", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-04-29", "release_date_basis": "announcement page", "weights_date": "2022-03-02", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta self-supervised vision transformer trained by self-distillation without labels; predecessor of DINOv2 and DINOv3.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2104.14294", "https://huggingface.co/facebook/dino-vitb16", "https://huggingface.co/api/models?pipeline_tag=image-feature-extraction&sort=downloads&limit=50", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md" ], "unverified": [] }, { "id": "mae", "name": "MAE (Masked Autoencoder)", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-11-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Self-supervised ViT pretraining that reconstructs masked image patches, from the paper \"Masked Autoencoders Are Scalable Vision Learners\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2111.06377", "https://github.com/open-mmlab/mmsegmentation/tree/main/configs/mae", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://github.com/tensorflow/models/tree/master/official/projects/mae", "https://arxiv.org/html/2111.06377" ], "unverified": [] }, { "id": "moco", "name": "MoCo", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "MoCo v1, v2 (arXiv 2003.04297) and v3 (arXiv 2104.02057).", "context_length": null, "release_date": "2019-11-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Self-supervised visual representation learning with momentum contrast; v3 extends it to vision transformers.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1911.05722", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/mocov2", "https://arxiv.org/abs/2003.04297", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/mocov3", "https://arxiv.org/abs/2104.02057", "https://github.com/facebookresearch/moco/tree/master/detection" ], "unverified": [] }, { "id": "google-simclr", "name": "SimCLR", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "SimCLR v1 and v2 in the TensorFlow Model Garden.", "context_length": null, "release_date": "2020-02-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google framework for contrastive self-supervised learning of visual representations.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2002.05709", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/simclr", "https://github.com/tensorflow/models/tree/master/official/projects/simclr" ], "unverified": [] }, { "id": "byol", "name": "BYOL", "level": "model", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-06-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Self-supervised learning without negative pairs, from the paper \"Bootstrap your own latent\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2006.07733", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/byol", "https://arxiv.org/html/2006.07733" ], "unverified": [] }, { "id": "swav", "name": "SwAV", "level": "model", "family_id": null, "developer": [ "inria", "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-06-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Self-supervised learning by contrasting cluster assignments.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2006.09882", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/swav", "https://arxiv.org/html/2006.09882" ], "unverified": [] }, { "id": "barlow-twins", "name": "Barlow Twins", "level": "model", "family_id": null, "developer": [ "meta", "nyu" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-03-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Self-supervised learning by redundancy reduction.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.03230", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/barlowtwins", "https://arxiv.org/html/2103.03230" ], "unverified": [] }, { "id": "i-jepa", "name": "I-JEPA", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-01-19", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Joint-embedding predictive architecture that learns image representations by predicting masked regions in feature space.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2301.08243", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/html/2301.08243" ], "unverified": [] }, { "id": "simmim", "name": "SimMIM", "level": "model", "family_id": null, "developer": [ "tsinghua-university", "microsoft-research" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-11-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Simple masked image modelling framework.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2111.09886", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/simmim", "https://arxiv.org/html/2111.09886" ], "unverified": [] }, { "id": "meta-data2vec-vision", "name": "data2vec vision", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-02-07", "release_date_basis": "announcement page", "weights_date": "2022-04-14", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Image model from Meta's data2vec, a self-supervised framework shared across speech, vision and language.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2202.03555", "https://huggingface.co/facebook/data2vec-vision-base", "https://ai.azure.com/catalog/models?task=image-classification" ], "unverified": [] }, { "id": "google-deepmind-perceiver", "name": "Perceiver (vision)", "level": "model", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-07-30", "release_date_basis": "announcement page", "weights_date": "2022-03-02", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DeepMind general architecture for structured inputs and outputs, published as ImageNet classifiers on Hugging Face.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2107.14795", "https://huggingface.co/deepmind/vision-perceiver-fourier", "https://ai.azure.com/catalog/models?task=image-classification", "https://github.com/tensorflow/models/tree/master/official/projects/perceiver" ], "unverified": [] }, { "id": "r-cnn", "name": "R-CNN family", "level": "family", "family_id": null, "developer": [ "uc-berkeley" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "R-CNN, Fast R-CNN, Faster R-CNN, Mask R-CNN, Cascade R-CNN, Keypoint R-CNN and their many backbones (ResNet, ResNeXt, Inception-ResNet, MobileNet v3).", "context_length": null, "release_date": "2013-11-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Region-based convolutional detector line from Ross Girshick and colleagues; Faster R-CNN and Mask R-CNN remain standard baselines in every detection model zoo.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1311.2524", "https://github.com/onnx/models/tree/main/validated/vision/classification/rcnn_ilsvrc13", "https://arxiv.org/html/1311.2524" ], "unverified": [] }, { "id": "r-cnn-original", "name": "R-CNN", "level": "variant", "family_id": "r-cnn", "developer": [ "uc-berkeley" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2013-11-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "The original region-based CNN detector, from the paper \"Rich feature hierarchies for accurate object detection and semantic segmentation\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1311.2524", "https://github.com/onnx/models/tree/main/validated/vision/classification/rcnn_ilsvrc13", "https://arxiv.org/html/1311.2524" ], "unverified": [] }, { "id": "fast-r-cnn", "name": "Fast R-CNN", "level": "variant", "family_id": "r-cnn", "developer": [ "microsoft-research" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2015-04-30", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Faster training and inference version of R-CNN that pools region features from one shared feature map.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1504.08083", "https://github.com/open-mmlab/mmdetection/tree/main/configs/fast_rcnn", "https://github.com/facebookresearch/detectron2/blob/main/MODEL_ZOO.md", "https://arxiv.org/html/1504.08083" ], "unverified": [] }, { "id": "faster-r-cnn", "name": "Faster R-CNN", "level": "variant", "family_id": "r-cnn", "developer": [ "microsoft-research", "university-of-science-and-technology-of-china" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Backbones R50, R101, X101 (C4, DC5, FPN), Inception-ResNet v2 (Open Images V4), MobileNet v3 (TorchVision).", "context_length": null, "release_date": "2015-06-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Two-stage object detector with a region proposal network; a standard baseline for detection in industry and research.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1506.01497", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/faster_rcnn_inception_resnet_v2_atrous_coco", "https://modelzoo.co/model/faster-r-cnn-caffe", "https://modelzoo.co/model/faster-rcnnpytorch", "https://modelzoo.co/model/simple-faster-rcnn-pytorch", "https://github.com/rbgirshick/py-faster-rcnn", "https://arxiv.org/html/1506.01497" ], "unverified": [] }, { "id": "mask-r-cnn", "name": "Mask R-CNN", "level": "variant", "family_id": "r-cnn", "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "R50, R101, X101, X152, LVIS long-tail, Swin-T and LSJ schedules in Detectron2, MMDetection, TorchVision, TF Model Garden and NVIDIA DLE.", "context_length": null, "release_date": "2017-03-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "mobile or NPU", "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Extends Faster R-CNN with a mask branch for instance segmentation; the reference instance segmentation model in Detectron2 and MMDetection.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1703.06870", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://aihub.qualcomm.com/models/maskrcnn", "https://github.com/pytorch/vision", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/mask_rcnn_inception_resnet_v2_atrous_coco", "https://huggingface.co/OpenVINO/maskrcnn_r50-fp16-ov", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/dle/models/maskrcnn__pyt_ckpt", "https://arxiv.org/html/1703.06870" ], "unverified": [] }, { "id": "keypoint-r-cnn", "name": "Keypoint R-CNN", "level": "variant", "family_id": "r-cnn", "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-03-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Mask R-CNN with a keypoint head for human pose estimation, shipped in Detectron2 and TorchVision.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1703.06870", "https://github.com/facebookresearch/detectron2/blob/main/MODEL_ZOO.md", "https://pytorch.org/vision/stable/models/keypoint_rcnn.html" ], "unverified": [] }, { "id": "cascade-r-cnn", "name": "Cascade R-CNN", "level": "variant", "family_id": "r-cnn", "developer": [ "uc-san-diego" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Cascade R-CNN and Cascade Mask R-CNN; Cascade RCNN-RS (arXiv 2107.00057).", "context_length": null, "release_date": "2017-12-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multi-stage detector trained with increasing IoU thresholds for higher-quality boxes, from the paper \"Cascade R-CNN\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1712.00726", "https://paperswithcode.com/method/cascade-r-cnn", "https://paperswithcode.com/method/cascade-mask-r-cnn", "https://huggingface.co/datasets/pwc-archive/methods", "http://arxiv.org/abs/1712.00726v1", "https://github.com/open-mmlab/mmdetection/tree/main/configs/cascade_rcnn", "https://github.com/facebookresearch/detectron2/blob/main/MODEL_ZOO.md", "https://arxiv.org/html/1712.00726" ], "unverified": [] }, { "id": "r-fcn", "name": "R-FCN", "level": "model", "family_id": null, "developer": [ "microsoft-research" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Region-based fully convolutional detector; the OpenVINO Open Model Zoo ships an R-FCN ResNet-101 COCO model.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/rfcn-resnet101-coco-tf", "https://arxiv.org/html/1605.06409" ], "unverified": [] }, { "id": "htc", "name": "Hybrid Task Cascade (HTC)", "level": "model", "family_id": null, "developer": [ "cuhk", "sensetime" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-01-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Instance segmentation model that interleaves detection and mask branches in a cascade.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1901.07518", "https://github.com/open-mmlab/mmdetection/tree/main/configs/htc", "https://arxiv.org/html/1901.07518" ], "unverified": [] }, { "id": "detectors", "name": "DetectoRS", "level": "model", "family_id": null, "developer": [ "jhu", "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-06-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Detector with recursive feature pyramid and switchable atrous convolution.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2006.02334", "https://github.com/open-mmlab/mmdetection/tree/main/configs/detectors", "https://arxiv.org/html/2006.02334" ], "unverified": [] }, { "id": "grid-r-cnn", "name": "Grid R-CNN", "level": "model", "family_id": null, "developer": [ "sensetime" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-11-29", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Two-stage detector that localises boxes with a grid point guided mechanism.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1811.12030", "https://github.com/open-mmlab/mmdetection/tree/main/configs/grid_rcnn", "https://arxiv.org/html/1811.12030" ], "unverified": [] }, { "id": "tridentnet", "name": "TridentNet", "level": "model", "family_id": null, "developer": [ "university-of-chinese-academy-of-sciences", "tusimple" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-01-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Scale-aware trident network for object detection with parallel dilated branches.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1901.01892", "https://github.com/open-mmlab/mmdetection/tree/main/configs/tridentnet", "https://github.com/facebookresearch/detectron2/blob/main/projects/README.md", "https://arxiv.org/html/1901.01892" ], "unverified": [] }, { "id": "sparse-r-cnn", "name": "Sparse R-CNN", "level": "model", "family_id": null, "developer": [ "hku", "bytedance" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-11-25", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "End-to-end detector with a small set of learnable proposals.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2011.12450", "https://github.com/open-mmlab/mmdetection/tree/main/configs/sparse_rcnn", "https://github.com/PeizeSun/SparseR-CNN", "https://arxiv.org/html/2011.12450" ], "unverified": [] }, { "id": "queryinst", "name": "QueryInst", "level": "model", "family_id": null, "developer": [ "hust" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-05-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Query-based instance segmentation, from the paper \"Instances as Queries\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2105.01928", "https://github.com/open-mmlab/mmdetection/tree/main/configs/queryinst", "https://arxiv.org/html/2105.01928" ], "unverified": [] }, { "id": "meta-detectron2", "name": "Detectron2 model zoo", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-09-05", "release_date_basis": "registry date", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta (Facebook AI Research) detection and segmentation library with a model zoo of Faster R-CNN, Mask R-CNN, RetinaNet, Keypoint R-CNN and panoptic models; deployable from Google Vertex AI.", "verified_on": "2026-10-12", "sources": [ "https://github.com/facebookresearch/detectron2", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://aihub.qualcomm.com/models/detectron2_detection", "https://github.com/facebookresearch/detectron2/", "https://modelzoo.co/model/detectron2" ], "unverified": [ "release_date" ] }, { "id": "meta-d2go", "name": "D2Go", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-02-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta toolkit and model zoo of mobile Faster R-CNN, Mask R-CNN and keypoint models built on Detectron2.", "verified_on": "2026-10-12", "sources": [ "https://github.com/facebookresearch/d2go" ], "unverified": [ "release_date" ] }, { "id": "meta-densepose", "name": "DensePose", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-02-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Dense human pose estimation that maps every human pixel to a 3D body surface, from Meta (Facebook AI Research).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1802.00434", "https://github.com/facebookresearch/DensePose", "https://modelzoo.co/model/densepose", "https://github.com/facebookresearch/detectron2/blob/main/projects/README.md" ], "unverified": [] }, { "id": "meta-pointrend", "name": "PointRend", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-12-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Instance and semantic segmentation that refines masks at adaptively chosen points, treating segmentation as rendering; in the Detectron2 project list.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1912.08193", "https://github.com/open-mmlab/mmdetection/tree/main/configs/point_rend", "https://github.com/facebookresearch/detectron2/blob/main/projects/README.md" ], "unverified": [] }, { "id": "meta-tensormask", "name": "TensorMask", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-03-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Dense sliding-window instance segmentation, from the Detectron2 project list.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1903.12174", "https://github.com/facebookresearch/detectron2/blob/main/projects/README.md" ], "unverified": [] }, { "id": "meta-panoptic-fpn", "name": "Panoptic FPN", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-01-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Panoptic segmentation that adds a semantic branch to Mask R-CNN, from the paper \"Panoptic Feature Pyramid Networks\"; Semantic FPN is its semantic part.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1901.02446", "https://github.com/open-mmlab/mmdetection/tree/main/configs/panoptic_fpn", "https://github.com/open-mmlab/mmsegmentation/tree/main/configs/sem_fpn", "https://github.com/facebookresearch/detectron2/blob/main/MODEL_ZOO.md" ], "unverified": [] }, { "id": "meta-vitdet", "name": "ViTDet", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-03-30", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Object detection with a plain, non-hierarchical ViT backbone, from the paper \"Exploring Plain Vision Transformer Backbones for Object Detection\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2203.16527", "https://github.com/open-mmlab/mmdetection/tree/main/projects/ViTDet", "https://github.com/facebookresearch/detectron2/blob/main/projects/README.md", "https://www.kaggle.com/models/keras/vitdet" ], "unverified": [] }, { "id": "meta-mvit", "name": "MViT (Multiscale Vision Transformers)", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "MViT v1 and MViTv2; image and video checkpoints.", "context_length": null, "release_date": "2021-04-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multiscale vision transformers for image classification, detection and video recognition, in Detectron2, PySlowFast and TorchVision.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2112.01526", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/mvit", "https://github.com/open-mmlab/mmaction2/tree/main/configs/recognition/mvit", "https://github.com/facebookresearch/detectron2/blob/main/projects/README.md", "https://pytorch.org/vision/stable/models/video_mvit.html", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/abs/2104.11227" ], "unverified": [] }, { "id": "centermask", "name": "CenterMask", "level": "model", "family_id": null, "developer": [ "etri" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-11-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time anchor-free instance segmentation built on FCOS with a VoVNetV2 backbone.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1911.06667", "https://github.com/youngwanLEE/centermask2", "https://arxiv.org/html/1911.06667" ], "unverified": [] }, { "id": "bcnet", "name": "BCNet", "level": "model", "family_id": null, "developer": [ "hkust", "kuaishou" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-03-23", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Occlusion-aware instance segmentation with overlapping bilayers.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.12340", "https://github.com/lkeab/BCNet", "https://arxiv.org/html/2103.12340" ], "unverified": [] }, { "id": "scnet", "name": "SCNet (instance segmentation)", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-12-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Instance segmentation model that keeps training and inference samples consistent.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2012.10150", "https://github.com/open-mmlab/mmdetection/tree/main/configs/scnet" ], "unverified": [] }, { "id": "yolact", "name": "YOLACT", "level": "model", "family_id": null, "developer": [ "uc-davis" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "YOLACT ResNet50 FPN (OpenVINO) and YOLACT-Edge (NXP).", "context_length": null, "release_date": "2019-04-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server", "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO, NXP eIQ model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time instance segmentation from the paper \"YOLACT: Real-time Instance Segmentation\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1904.02689", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/yolact-resnet50-fpn-pytorch", "https://github.com/NXP/eiq-model-zoo/tree/main/tasks/vision/instance-segmentation", "https://github.com/open-mmlab/mmdetection/tree/main/configs/yolact", "https://arxiv.org/html/1904.02689" ], "unverified": [] }, { "id": "solo", "name": "SOLO", "level": "model", "family_id": null, "developer": [ "university-of-adelaide", "bytedance" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "SOLO and SOLOv2 (arXiv 2003.10152).", "context_length": null, "release_date": "2019-12-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Box-free instance segmentation that segments objects by location.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1912.04488", "https://github.com/open-mmlab/mmdetection/tree/main/configs/solo", "https://github.com/open-mmlab/mmdetection/tree/main/configs/solov2", "https://arxiv.org/abs/2003.10152", "https://arxiv.org/html/1912.04488" ], "unverified": [] }, { "id": "condinst", "name": "CondInst", "level": "model", "family_id": null, "developer": [ "university-of-adelaide" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-03-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Instance segmentation with dynamic conditional convolutions.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2003.05664", "https://github.com/open-mmlab/mmdetection/tree/main/configs/condinst", "https://arxiv.org/html/2003.05664" ], "unverified": [] }, { "id": "blendmask", "name": "BlendMask", "level": "model", "family_id": null, "developer": [ "university-of-adelaide" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-01-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Instance segmentation that blends top-down and bottom-up information, in the AdelaiDet toolbox with FCOS.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2001.00309", "https://github.com/aim-uofa/adet", "https://arxiv.org/html/2001.00309" ], "unverified": [] }, { "id": "sparseinst", "name": "SparseInst", "level": "model", "family_id": null, "developer": [ "hust" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-03-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time instance segmentation with sparse instance activation maps.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2203.12827", "https://github.com/hustvl/SparseInst", "https://github.com/open-mmlab/mmdetection/tree/main/projects/SparseInst" ], "unverified": [] }, { "id": "knet", "name": "K-Net", "level": "model", "family_id": null, "developer": [ "ntu-singapore", "sensetime" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-06-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Unified semantic, instance and panoptic segmentation with learnable kernels.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2106.14855", "https://github.com/open-mmlab/mmsegmentation/tree/main/configs/knet", "https://arxiv.org/html/2106.14855" ], "unverified": [] }, { "id": "retinanet", "name": "RetinaNet", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "R50 and R101 FPN in Detectron2, MMDetection, TorchVision, TF Model Garden; RetinaNet-RS.", "context_length": null, "release_date": "2017-08-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "One-stage detector trained with focal loss, from the paper \"Focal Loss for Dense Object Detection\"; a standard dense detector baseline.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1708.02002", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/retinanet-tf", "https://modelzoo.co/model/pytorch-retinanet-2", "https://github.com/yhenon/pytorch-retinanet", "https://www.kaggle.com/models/tensorflow/retinanet-resnet", "https://www.kaggle.com/models/keras/retinanet", "https://arxiv.org/html/1708.02002" ], "unverified": [] }, { "id": "fcos", "name": "FCOS", "level": "model", "family_id": null, "developer": [ "university-of-adelaide" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-04-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Anchor-free one-stage detector, from the paper \"FCOS: Fully Convolutional One-Stage Object Detection\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1904.01355", "https://paperswithcode.com/method/fcos", "https://huggingface.co/datasets/pwc-archive/methods", "https://arxiv.org/abs/1904.01355v5", "https://github.com/open-mmlab/mmdetection/tree/main/configs/fcos", "https://github.com/aim-uofa/adet", "https://arxiv.org/abs/2001.00309", "https://arxiv.org/html/1904.01355" ], "unverified": [] }, { "id": "nas-fcos", "name": "NAS-FCOS", "level": "model", "family_id": null, "developer": [ "nwpu", "university-of-adelaide" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-06-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "FCOS detector whose head and neck were found by neural architecture search.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1906.04423", "https://github.com/open-mmlab/mmdetection/tree/main/configs/nas_fcos", "https://arxiv.org/html/1906.04423" ], "unverified": [] }, { "id": "nas-fpn", "name": "NAS-FPN", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-04-16", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "RetinaNet with a feature pyramid found by neural architecture search.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1904.07392", "https://github.com/open-mmlab/mmdetection/tree/main/configs/nas_fpn", "https://arxiv.org/html/1904.07392" ], "unverified": [] }, { "id": "spinenet", "name": "SpineNet", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-12-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Scale-permuted backbone for recognition and localisation, in the TensorFlow Model Garden.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1912.05027", "https://github.com/tensorflow/models/blob/master/official/README.md", "https://arxiv.org/html/1912.05027" ], "unverified": [] }, { "id": "foveabox", "name": "FoveaBox", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-04-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Anchor-free detector, from the paper \"FoveaBox: Beyond Anchor-based Object Detector\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1904.03797", "https://github.com/open-mmlab/mmdetection/tree/main/configs/foveabox" ], "unverified": [] }, { "id": "reppoints", "name": "RepPoints", "level": "model", "family_id": null, "developer": [ "peking-university", "microsoft-research" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-04-25", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Detector that represents objects as sets of points, from the paper \"RepPoints\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1904.11490", "https://github.com/open-mmlab/mmdetection/tree/main/configs/reppoints", "https://arxiv.org/html/1904.11490" ], "unverified": [] }, { "id": "tood", "name": "TOOD", "level": "model", "family_id": null, "developer": [ "intellifusion" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-08-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Task-aligned one-stage object detector.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2108.07755", "https://github.com/open-mmlab/mmdetection/tree/main/configs/tood", "https://arxiv.org/html/2108.07755" ], "unverified": [] }, { "id": "cornernet", "name": "CornerNet", "level": "model", "family_id": null, "developer": [ "princeton-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-08-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Detector that finds objects as paired top-left and bottom-right keypoints.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1808.01244", "https://github.com/open-mmlab/mmdetection/tree/main/configs/cornernet", "https://arxiv.org/html/1808.01244" ], "unverified": [] }, { "id": "centripetalnet", "name": "CentripetalNet", "level": "model", "family_id": null, "developer": [ "xi-an-jiaotong-university", "sensetime" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-03-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Keypoint-based detector that pairs corners with centripetal shifts.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2003.09119", "https://github.com/open-mmlab/mmdetection/tree/main/configs/centripetalnet", "https://arxiv.org/html/2003.09119" ], "unverified": [] }, { "id": "centernet", "name": "CenterNet (Objects as Points)", "level": "model", "family_id": null, "developer": [ "ut-austin", "uc-berkeley" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Hourglass, ResNet, DLA-34 and MobileNetV2 FPN backbones; 2D boxes, 3D boxes and human pose.", "context_length": null, "release_date": "2019-04-16", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "mobile or NPU", "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, OpenVINO, NXP eIQ model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Detector that models objects as their centre points and regresses size, 3D box and pose from them; in TF Model Garden, OpenVINO and NXP zoos.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1904.07850", "https://github.com/xingyizhou/CenterNet", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://aihub.qualcomm.com/models/centernet_2d", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/ctdet_coco_dlav0_512", "https://github.com/NXP/eiq-model-zoo/tree/main/tasks/vision/object-detection", "https://www.kaggle.com/models/tensorflow/centernet-hourglass", "https://arxiv.org/html/1904.07850" ], "unverified": [] }, { "id": "extremenet", "name": "ExtremeNet", "level": "model", "family_id": null, "developer": [ "ut-austin" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Bottom-up detector that groups extreme points; listed in the TensorFlow 2 Detection Model Zoo.", "verified_on": "2026-10-12", "sources": [ "https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/tf2_detection_zoo.md", "https://arxiv.org/html/1901.08043" ], "unverified": [] }, { "id": "ssd", "name": "SSD (Single Shot MultiBox Detector)", "level": "family", "family_id": null, "developer": [ "unc-chapel-hill", "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "SSD300 and SSD512 with VGG16, ResNet-34 (MLPerf) and ResNet-50 (NVIDIA); SSD MobileNet v1 is already in AURION (ssd-mobilenet-v1).", "context_length": null, "release_date": "2015-12-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "mobile or NPU", "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, OpenVINO, TI edge AI model zoo, MediaPipe.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "One-stage multi-box detector from Wei Liu and colleagues; with MobileNet backbones it is the standard detector on phones and microcontrollers.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1512.02325", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://aihub.qualcomm.com/models/resnet34_ssd1200", "https://github.com/mlcommons/inference/tree/33894a19c4af6207f7cfdda75f84570f04836de5/vision/classification_and_detection", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/ssd-resnet34-1200-onnx", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/dle/models/ssd__pyt_ckpt", "https://github.com/TexasInstruments/edgeai-tensorlab/tree/main/edgeai-modelzoo/models/vision/detection/coco", "https://arxiv.org/html/1512.02325" ], "unverified": [] }, { "id": "ssd-mobilenet-v2", "name": "SSD MobileNet v2", "level": "variant", "family_id": "ssd", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "SSD MobileNet v2 and v2 FPNLite (320, 640).", "context_length": null, "release_date": "2018-01-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: MediaPipe.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "SSD detector with a MobileNet v2 backbone, introduced in the MobileNetV2 paper; widely used on TensorFlow Lite and NPUs.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1801.04381", "https://ai.google.dev/edge/mediapipe/solutions/vision/object_detector", "https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/tf2_detection_zoo.md", "https://www.kaggle.com/models/tensorflow/ssd-mobilenet-v2", "https://www.kaggle.com/models/google/mobilenet-v2", "https://arxiv.org/html/1801.04381" ], "unverified": [] }, { "id": "ssdlite", "name": "SSDLite", "level": "variant", "family_id": "ssd", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-01-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "SSD with depthwise separable prediction heads, introduced in the MobileNetV2 paper and later used with MobileNetV3.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1905.02244", "https://pytorch.org/vision/stable/models/ssdlite.html", "https://www.kaggle.com/models/tensorflow/ssdlite-mobilenet-v2", "https://arxiv.org/html/1905.02244", "https://arxiv.org/abs/1801.04381" ], "unverified": [] }, { "id": "ssd-resnet34", "name": "SSD-ResNet34 (MLPerf)", "level": "variant", "family_id": "ssd", "developer": [ "unc-chapel-hill", "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2015-12-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU", "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "SSD with a ResNet-34 backbone at 1200x1200, the MLPerf Inference reference detector.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1512.02325", "https://github.com/mlcommons/inference", "https://aihub.qualcomm.com/models/resnet34_ssd1200", "https://github.com/mlcommons/inference/tree/33894a19c4af6207f7cfdda75f84570f04836de5/vision/classification_and_detection", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/ssd-resnet34-1200-onnx", "https://arxiv.org/html/1512.02325" ], "unverified": [] }, { "id": "nvidia-ssd300", "name": "NVIDIA SSD300", "level": "variant", "family_id": "ssd", "developer": [ "nvidia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA SSD300 detector with a ResNet-50 backbone, optimised for NVIDIA GPUs in Deep Learning Examples.", "verified_on": "2026-10-12", "sources": [ "https://github.com/NVIDIA/DeepLearningExamples/tree/master/PyTorch/Detection/SSD", "https://github.com/NVIDIA/DeepLearningExamples", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/dle/models/ssd__pyt_ckpt", "https://pytorch.org/hub/nvidia_deeplearningexamples_ssd", "https://raw.githubusercontent.com/pytorch/hub/master/nvidia_deeplearningexamples_ssd.md" ], "unverified": [] }, { "id": "google-efficientdet", "name": "EfficientDet", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "D0 to D7 and EfficientDet-Lite0 to Lite4.", "context_length": null, "release_date": "2019-11-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "CPU server", "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO, NXP eIQ model zoo, TI edge AI model zoo, MediaPipe.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google family of scalable detectors combining an EfficientNet backbone with a BiFPN feature network; Lite versions target phones and edge NPUs.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1911.09070", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/efficientdet-d0-tf", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/dle/models/efficientdet__tf2_ckpt", "https://github.com/NXP/eiq-model-zoo/tree/main/tasks/vision/object-detection", "https://github.com/TexasInstruments/edgeai-tensorlab/tree/main/edgeai-modelzoo/models/vision/detection/coco", "https://modelzoo.co/model/yet-another-efficientdet-pytorch" ], "unverified": [] }, { "id": "google-efficientdet-d0", "name": "EfficientDet-D0", "level": "variant", "family_id": "google-efficientdet", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": 3900000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-11-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "The smallest standard EfficientDet, about 3.9M parameters.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1911.09070" ], "unverified": [] }, { "id": "google-efficientdet-d7", "name": "EfficientDet-D7", "level": "variant", "family_id": "google-efficientdet", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-11-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "The largest EfficientDet in the original paper.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1911.09070" ], "unverified": [] }, { "id": "google-efficientdet-lite", "name": "EfficientDet-Lite", "level": "variant", "family_id": "google-efficientdet", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Lite0 to Lite4.", "context_length": null, "release_date": "2019-11-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: NXP eIQ model zoo, MediaPipe.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Mobile versions of EfficientDet for TensorFlow Lite, MediaPipe and edge NPUs.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1911.09070", "https://github.com/NXP/eiq-model-zoo/tree/main/tasks/vision/object-detection", "https://ai.google.dev/edge/mediapipe/solutions/vision/object_detector" ], "unverified": [] }, { "id": "google-mosaic", "name": "MOSAIC (mobile segmentation)", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google mobile semantic segmentation model published on Kaggle Models and in the TensorFlow Model Garden.", "verified_on": "2026-10-12", "sources": [ "https://www.kaggle.com/models/google/mosaic", "https://github.com/tensorflow/models/tree/master/official/projects/mosaic" ], "unverified": [] }, { "id": "google-lstm-mobile-video-detector", "name": "LSTM mobile video object detector", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-11-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Mobile video object detector with temporally aware feature maps, from the paper \"Mobile Video Object Detection with Temporally-Aware Feature Maps\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1711.06368", "https://github.com/tensorflow/models/blob/master/research/README.md" ], "unverified": [] }, { "id": "google-pix2seq", "name": "Pix2Seq", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google detector that outputs boxes as a sequence of tokens, in the TensorFlow Model Garden.", "verified_on": "2026-10-12", "sources": [ "https://github.com/tensorflow/models/tree/master/official/projects/pix2seq" ], "unverified": [] }, { "id": "google-deepmac", "name": "DeepMAC Mask R-CNN", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google Mask R-CNN variant whose mask head generalises to classes without mask labels, in the TensorFlow Model Garden.", "verified_on": "2026-10-12", "sources": [ "https://github.com/tensorflow/models/tree/master/official/projects/deepmac_maskrcnn" ], "unverified": [] }, { "id": "google-panoptic-deeplab", "name": "Panoptic-DeepLab", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-11-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Bottom-up panoptic segmentation baseline; the TensorFlow Model Garden ships it with Panoptic Mask R-CNN.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1911.10194", "https://github.com/facebookresearch/detectron2/blob/main/projects/README.md", "https://github.com/tensorflow/models/tree/master/official/projects/panoptic" ], "unverified": [] }, { "id": "yolo", "name": "YOLO (original line, v1 to v3)", "level": "family", "family_id": null, "developer": [ "university-of-washington", "allen-ai" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "YOLOv1 and tiny, YOLOv2 (YOLO9000) and Tiny YOLOv2, YOLOv3; YOLOv3-tiny is already in AURION (yolo-v3-tiny).", "context_length": null, "release_date": "2015-06-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server", "microcontroller" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO, STM32 model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Joseph Redmon's single-pass real-time detectors that started the YOLO naming; later YOLO versions come from other teams.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1506.02640", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/yolo-v1-tiny-tf", "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/object_detection/yolov2t", "https://github.com/onnx/models/tree/main/validated/vision/object_detection_segmentation/tiny-yolov2", "https://arxiv.org/abs/1612.08242", "https://github.com/onnx/models/tree/main/validated/vision/object_detection_segmentation/yolov2-coco", "https://github.com/onnx/models/tree/main/validated/vision/object_detection_segmentation/yolov3", "https://arxiv.org/html/1506.02640" ], "unverified": [] }, { "id": "yolo-v1", "name": "YOLOv1", "level": "variant", "family_id": "yolo", "developer": [ "university-of-washington", "allen-ai" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2015-06-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "The first YOLO, from the paper \"You Only Look Once: Unified, Real-Time Object Detection\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1506.02640", "https://arxiv.org/html/1506.02640" ], "unverified": [] }, { "id": "yolo-v2", "name": "YOLOv2 (YOLO9000)", "level": "variant", "family_id": "yolo", "developer": [ "university-of-washington", "allen-ai" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "YOLOv2 and Tiny YOLOv2 (ST STM32 port).", "context_length": null, "release_date": "2016-12-25", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "microcontroller" ], "footprint_note": "Edge or toolkit builds listed in: STM32 model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Second YOLO, from the paper \"YOLO9000: Better, Faster, Stronger\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1612.08242", "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/object_detection/yolov2t", "https://github.com/onnx/models/tree/main/validated/vision/object_detection_segmentation/tiny-yolov2", "https://github.com/onnx/models/tree/main/validated/vision/object_detection_segmentation/yolov2-coco", "https://arxiv.org/html/1612.08242" ], "unverified": [] }, { "id": "yolo-v3", "name": "YOLOv3", "level": "variant", "family_id": "yolo", "developer": [ "university-of-washington" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-04-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Third YOLO with a DarkNet-53 backbone, from the paper \"YOLOv3: An Incremental Improvement\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1804.02767", "https://github.com/onnx/models/tree/main/validated/vision/object_detection_segmentation/yolov3", "https://arxiv.org/html/1804.02767" ], "unverified": [] }, { "id": "yolo-v4", "name": "YOLOv4", "level": "model", "family_id": null, "developer": [ "academia-sinica" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "YOLOv4, YOLOv4-tiny (NXP, OpenVINO) and Scaled-YOLOv4.", "context_length": null, "release_date": "2020-04-23", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU", "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, OpenVINO, NXP eIQ model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time detector by Bochkovskiy, Wang and Liao with a CSPDarkNet-53 backbone, from the paper \"YOLOv4: Optimal Speed and Accuracy of Object Detection\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2004.10934", "https://aihub.qualcomm.com/models/yolov4", "https://github.com/bubbliiiing/yolov4-tiny-pytorch.git", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/mobilenet-yolo-v4-syg", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/yolo-v4-tf", "https://github.com/NXP/eiq-model-zoo/tree/main/tasks/vision/object-detection", "https://modelzoo.co/model/pytorch-yolov4", "https://arxiv.org/html/2004.10934" ], "unverified": [] }, { "id": "yolor", "name": "YOLOR", "level": "model", "family_id": null, "developer": [ "academia-sinica" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-05-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "gpl-3-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Detector with implicit and explicit knowledge, from the paper \"You Only Learn One Representation\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2105.04206", "https://github.com/WongKinYiu/yolor", "https://aihub.qualcomm.com/models/yolor", "https://modelzoo.co/model/yolor", "https://arxiv.org/html/2105.04206" ], "unverified": [] }, { "id": "yolo-v7", "name": "YOLOv7", "level": "model", "family_id": null, "developer": [ "academia-sinica" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-07-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "gpl-3-0", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, TI edge AI model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time detector by Wang, Bochkovskiy and Liao, from the paper \"YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2207.02696", "https://github.com/WongKinYiu/yolov7", "https://aihub.qualcomm.com/models/yolov7", "https://github.com/WongKinYiu/yolov7/", "https://github.com/TexasInstruments/edgeai-tensorlab/tree/main/edgeai-modelzoo/models/vision/detection/coco", "https://docs.roboflow.com/models/supported-models/yolov7.md", "https://modelzoo.co/model/yolov7", "https://arxiv.org/html/2207.02696" ], "unverified": [] }, { "id": "yolo-v9", "name": "YOLOv9", "level": "model", "family_id": null, "developer": [ "academia-sinica" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-02-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "API" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time detector with programmable gradient information, from the paper \"YOLOv9: Learning What You Want to Learn Using Programmable Gradient Information\"; also served by Ultralytics and Roboflow.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2402.13616", "https://aihub.qualcomm.com/models/yolov9_det", "https://github.com/ultralytics/ultralytics/tree/main/ultralytics/models/yolo/detect", "https://docs.roboflow.com/models/supported-models/yolov9.md", "https://docs.ultralytics.com/models/yolov9/", "https://arxiv.org/html/2402.13616" ], "unverified": [] }, { "id": "meituan-yolov6", "name": "YOLOv6", "level": "model", "family_id": null, "developer": [ "meituan" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "YOLOv6 v1 to v3.0 (arXiv 2301.05586), N to L.", "context_length": null, "release_date": "2022-09-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "gpl-3-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meituan single-stage detector designed for industrial applications.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2209.02976", "https://github.com/meituan/YOLOv6", "https://aihub.qualcomm.com/models/yolov6", "https://github.com/meituan/YOLOv6/", "https://github.com/open-mmlab/mmyolo/tree/main/configs/yolov6", "https://docs.ultralytics.com/models/yolov6/", "https://arxiv.org/abs/2301.05586" ], "unverified": [] }, { "id": "megvii-yolox", "name": "YOLOX", "level": "model", "family_id": null, "developer": [ "megvii" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Nano, Tiny, S, M, L, X; TI lite variants and OpenVINO ports.", "context_length": null, "release_date": "2021-07-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU", "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, OpenVINO, TI edge AI model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Megvii anchor-free YOLO detector, from the paper \"YOLOX: Exceeding YOLO Series in 2021\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2107.08430", "https://github.com/Megvii-BaseDetection/YOLOX", "https://aihub.qualcomm.com/models/yolox", "https://github.com/Megvii-BaseDetection/YOLOX/", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/yolox-tiny", "https://huggingface.co/OpenVINO/yolox_tiny-fp16-ov", "https://github.com/TexasInstruments/edgeai-tensorlab/tree/main/edgeai-modelzoo/models/vision/detection/coco" ], "unverified": [] }, { "id": "yolof", "name": "YOLOF", "level": "model", "family_id": null, "developer": [ "casia", "megvii" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-03-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Detector that uses a single-level feature map, from the paper \"You Only Look One-level Feature\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.09460", "https://ai.azure.com/catalog/models?task=object-detection", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/yolof", "https://github.com/open-mmlab/mmdetection/tree/main/configs/yolof", "https://arxiv.org/html/2103.09460" ], "unverified": [] }, { "id": "pp-yoloe", "name": "PP-YOLOE", "level": "model", "family_id": null, "developer": [ "baidu" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-03-30", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Anchor-free YOLO detector from the PaddlePaddle team, from the paper \"PP-YOLOE: An evolved version of YOLO\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2203.16250", "https://github.com/open-mmlab/mmyolo/tree/main/configs/ppyoloe", "https://arxiv.org/html/2203.16250" ], "unverified": [] }, { "id": "alibaba-damo-damo-yolo", "name": "DAMO-YOLO", "level": "model", "family_id": null, "developer": [ "alibaba-damo" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-11-23", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba DAMO Academy real-time detector with a NAS-designed backbone, published on ModelScope.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2211.15444", "https://www.modelscope.cn/models/iic/cv_tinynas_object-detection_damoyolo" ], "unverified": [] }, { "id": "yolo-nas", "name": "YOLO-NAS", "level": "model", "family_id": null, "developer": [ "deci" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-05-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Detector found by neural architecture search and released by Deci AI in the SuperGradients library.", "verified_on": "2026-10-12", "sources": [ "https://github.com/Deci-AI/super-gradients", "https://universe.roboflow.com/search?q=sawit", "https://docs.ultralytics.com/models/yolo-nas/", "https://raw.githubusercontent.com/Deci-AI/super-gradients/HEAD/README.md", "https://github.com/Deci-AI/super-gradients/blob/master/LICENSE.YOLONAS.md" ], "unverified": [] }, { "id": "tsinghua-university-yoloe", "name": "YOLOE", "level": "model", "family_id": null, "developer": [ "tsinghua-university" ], "model_class": "vision", "modalities_in": [ "image", "text" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "YOLOE-v8, YOLOE-11 and YOLOE-26 sizes.", "context_length": null, "release_date": "2025-03-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "agpl-3-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open-vocabulary real-time detector and segmenter prompted by text, visual examples or nothing, from Tsinghua University.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2503.07465", "https://github.com/THU-MIG/yoloe", "https://aihub.qualcomm.com/models/yoloe_seg", "https://docs.ultralytics.com/models/yoloe/" ], "unverified": [] }, { "id": "tencent-yolo-world", "name": "YOLO-World", "level": "model", "family_id": null, "developer": [ "tencent" ], "model_class": "vision", "modalities_in": [ "image", "text" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "YOLO-World v1 and v2, S to X.", "context_length": null, "release_date": "2024-01-30", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "gpl-3-0", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tencent AI Lab real-time open-vocabulary detector that finds objects named in a text prompt.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2401.17270", "https://github.com/AILab-CVC/YOLO-World", "https://aihub.qualcomm.com/models/yolo_world", "https://replicate.com/zsxkib/yolo-world", "https://docs.roboflow.com/models/supported-models/yolo-world.md", "https://docs.ultralytics.com/models/yolo-world/" ], "unverified": [] }, { "id": "yolo-pose", "name": "YOLO-Pose", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-04-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multi-person pose estimation built into a YOLO detector, from the paper \"YOLO-Pose\"; TI ships it in its edge AI model zoo.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2204.06806", "https://github.com/open-mmlab/mmpose/tree/main/configs" ], "unverified": [] }, { "id": "hust-yolos", "name": "YOLOS", "level": "model", "family_id": null, "developer": [ "hust" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 30684768, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-06-01", "release_date_basis": "announcement page", "weights_date": "2022-04-26", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Plain ViT object detector, from the paper \"You Only Look at One Sequence\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2106.00666", "https://huggingface.co/hustvl/yolos-small", "https://ai.azure.com/catalog/models?task=object-detection", "https://huggingface.co/api/models?pipeline_tag=object-detection&sort=downloads&limit=50" ], "unverified": [] }, { "id": "yolov5-face", "name": "YOLOv5-Face", "level": "model", "family_id": null, "developer": [ "deepcam" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-04-26", "release_date_basis": "registry date", "weights_date": null, "openness": "open source", "licence": "gpl-3-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Face detector with landmarks built on YOLOv5.", "verified_on": "2026-10-12", "sources": [ "https://github.com/deepcam-cn/yolov5-face", "https://aihub.qualcomm.com/models/yolov5_face", "https://arxiv.org/html/2105.12931" ], "unverified": [ "release_date" ] }, { "id": "nanodet", "name": "NanoDet", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "NanoDet-m and NanoDet-Plus-m.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server", "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO, NXP eIQ model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Ultra-lightweight anchor-free detector for mobile devices.", "verified_on": "2026-10-12", "sources": [ "https://github.com/RangiLyu/nanodet", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/nanodet-m-1.5x-416", "https://github.com/NXP/eiq-model-zoo/tree/main/tasks/vision/object-detection" ], "unverified": [] }, { "id": "fastestdet", "name": "FastestDet", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: NXP eIQ model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Very small single-scale object detector; NXP ships a recipe for i.MX processors.", "verified_on": "2026-10-12", "sources": [ "https://github.com/NXP/eiq-model-zoo/tree/main/tasks/vision/object-detection" ], "unverified": [] }, { "id": "openmmlab-rtmdet", "name": "RTMDet", "level": "model", "family_id": null, "developer": [ "openmmlab" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Tiny, S, M, L, X; RTMDet-Ins and rotated variants.", "context_length": null, "release_date": "2022-12-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU", "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, OpenVINO, TI edge AI model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenMMLab real-time detector and instance segmenter, from the paper \"RTMDet: An Empirical Study of Designing Real-Time Object Detectors\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2212.07784", "https://aihub.qualcomm.com/models/rtmdet", "https://github.com/open-mmlab/mmdetection/tree/3.x/configs/rtmdet", "https://huggingface.co/OpenVINO/rtmdet_inst_tiny-fp16-ov", "https://github.com/TexasInstruments/edgeai-tensorlab/tree/main/edgeai-modelzoo/models/vision/detection/coco", "https://github.com/open-mmlab/mmdetection/tree/main/configs/rtmdet" ], "unverified": [] }, { "id": "meta-detr", "name": "DETR", "level": "family", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "DETR R50, R101 and DC5; panoptic heads.", "context_length": null, "release_date": "2020-05-26", "release_date_basis": "announcement page", "weights_date": "2022-03-02", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "mobile or NPU", "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, OpenVINO, Apple Core ML, TI edge AI model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta end-to-end transformer detector that predicts a set of boxes without anchors or non-maximum suppression; the base of RT-DETR and RF-DETR.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2005.12872", "https://huggingface.co/facebook/detr-resnet-50", "https://ai.azure.com/catalog/models?task=image-segmentation", "https://aihub.qualcomm.com/models/detr_resnet50", "https://github.com/facebookresearch/detr", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/detr-resnet50", "https://developer.apple.com/machine-learning/models/" ], "unverified": [] }, { "id": "meta-detr-resnet-50", "name": "DETR ResNet-50", "level": "variant", "family_id": "meta-detr", "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": 41000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-05-26", "release_date_basis": "announcement page", "weights_date": "2022-03-02", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DETR with a ResNet-50 backbone trained on COCO.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2005.12872", "https://huggingface.co/facebook/detr-resnet-50" ], "unverified": [] }, { "id": "meta-detr-resnet-101", "name": "DETR ResNet-101", "level": "variant", "family_id": "meta-detr", "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": 60675360, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-05-26", "release_date_basis": "announcement page", "weights_date": "2022-03-02", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DETR with a ResNet-101 backbone trained on COCO.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2005.12872", "https://huggingface.co/facebook/detr-resnet-101" ], "unverified": [] }, { "id": "meta-detr-resnet-50-panoptic", "name": "DETR ResNet-50 panoptic", "level": "variant", "family_id": "meta-detr", "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-05-26", "release_date_basis": "announcement page", "weights_date": "2022-03-02", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DETR with a panoptic segmentation head.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2005.12872", "https://huggingface.co/facebook/detr-resnet-50-panoptic" ], "unverified": [] }, { "id": "deformable-detr", "name": "Deformable DETR", "level": "model", "family_id": null, "developer": [ "sensetime", "university-of-science-and-technology-of-china" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 40175905, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-10-08", "release_date_basis": "announcement page", "weights_date": "2022-03-02", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DETR with deformable attention for faster convergence and small objects.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2010.04159", "https://huggingface.co/SenseTime/deformable-detr", "https://ai.azure.com/catalog/models?task=object-detection", "https://aihub.qualcomm.com/models/deformable_detr", "https://github.com/fundamentalvision/Deformable-DETR", "https://paperswithcode.com/method/deformable-detr", "https://huggingface.co/datasets/pwc-archive/methods", "https://arxiv.org/html/2010.04159" ], "unverified": [] }, { "id": "microsoft-conditional-detr", "name": "Conditional DETR", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 43524641, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-08-13", "release_date_basis": "announcement page", "weights_date": "2022-09-09", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DETR with conditional spatial queries for faster training.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2108.06152", "https://huggingface.co/microsoft/conditional-detr-resnet-50", "https://ai.azure.com/catalog/models?task=object-detection", "https://aihub.qualcomm.com/models/conditional_detr_resnet50", "https://github.com/huggingface/transformers/tree/main/src/transformers/models/conditional_detr", "https://github.com/open-mmlab/mmdetection/tree/main/configs/conditional_detr" ], "unverified": [ "developer" ] }, { "id": "idea-research-dino-detr", "name": "DINO (DETR with improved denoising anchor boxes)", "level": "family", "family_id": null, "developer": [ "idea-research" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "DAB-DETR (arXiv 2201.12329), DN-DETR (arXiv 2203.01305), DINO and Mask DINO.", "context_length": null, "release_date": "2022-03-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "IDEA Research line of DETR detectors with dynamic anchor boxes and query denoising, released in the detrex toolbox.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2203.03605", "https://github.com/IDEA-Research/detrex", "https://www.modelscope.cn/models/iic/cv_swinl_image-object-detection_dino", "https://github.com/open-mmlab/mmdetection/tree/main/configs/dab_detr", "https://arxiv.org/abs/2201.12329", "https://github.com/open-mmlab/mmdetection/tree/main/configs/dino", "https://arxiv.org/abs/2203.01305" ], "unverified": [] }, { "id": "idea-research-dab-detr", "name": "DAB-DETR", "level": "variant", "family_id": "idea-research-dino-detr", "developer": [ "idea-research" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-01-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "DETR with dynamic anchor boxes as queries.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2201.12329" ], "unverified": [] }, { "id": "idea-research-dn-detr", "name": "DN-DETR", "level": "variant", "family_id": "idea-research-dino-detr", "developer": [ "idea-research" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-03-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "DETR trained with query denoising.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2203.01305" ], "unverified": [] }, { "id": "idea-research-dino", "name": "DINO detector", "level": "variant", "family_id": "idea-research-dino-detr", "developer": [ "idea-research" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-03-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DETR with improved denoising anchor boxes; NVIDIA TAO also ships DINO detectors.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2203.03605", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/dino_with_fm_backbone" ], "unverified": [] }, { "id": "idea-research-mask-dino", "name": "Mask DINO", "level": "variant", "family_id": "idea-research-dino-detr", "developer": [ "idea-research" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-06-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Unified transformer for detection and segmentation built on DINO.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2206.02777", "https://www.modelscope.cn/models/iic/cv_maskdino-swin-l_image-instance-segmentation_coco" ], "unverified": [] }, { "id": "co-detr", "name": "Co-DETR", "level": "model", "family_id": null, "developer": [ "sensetime" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-11-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DETR trained with collaborative hybrid assignments.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2211.12860", "https://github.com/open-mmlab/mmdetection/tree/main/projects/CO-DETR", "https://arxiv.org/html/2211.12860" ], "unverified": [] }, { "id": "ddq", "name": "DDQ", "level": "model", "family_id": null, "developer": [ "hku", "shanghai-ai-lab" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-03-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "End-to-end detector with dense distinct queries.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2303.12776", "https://github.com/open-mmlab/mmdetection/tree/main/configs/ddq", "https://arxiv.org/html/2303.12776" ], "unverified": [] }, { "id": "diffusiondet", "name": "DiffusionDet", "level": "model", "family_id": null, "developer": [ "hku", "tencent" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-11-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Detector that denoises random boxes into object boxes, from the paper \"DiffusionDet: Diffusion Model for Object Detection\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2211.09788", "https://github.com/open-mmlab/mmdetection/tree/main/projects/DiffusionDet", "https://arxiv.org/html/2211.09788" ], "unverified": [] }, { "id": "idea-research-grounding-dino", "name": "Grounding DINO", "level": "model", "family_id": null, "developer": [ "idea-research" ], "model_class": "vision", "modalities_in": [ "image", "text" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 232810880, "params_active": null, "params_checkpoint": null, "params_note": "Tiny and Base; NVIDIA TAO Grounding DINO and Mask Grounding DINO builds.", "context_length": null, "release_date": "2023-03-09", "release_date_basis": "announcement page", "weights_date": "2023-09-25", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "IDEA Research open-set detector that finds objects described by text, used for auto-labelling and with SAM for segmentation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2303.05499", "https://huggingface.co/IDEA-Research/grounding-dino-base", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/grounding_dino", "https://huggingface.co/api/models?pipeline_tag=zero-shot-object-detection&sort=downloads&limit=50", "https://replicate.com/adirik/grounding-dino", "https://docs.roboflow.com/models/supported-models/grounding-dino.md" ], "unverified": [] }, { "id": "openmmlab-mm-grounding-dino", "name": "MM-Grounding-DINO", "level": "model", "family_id": null, "developer": [ "openmmlab" ], "model_class": "vision", "modalities_in": [ "image", "text" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 343550288, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-01-04", "release_date_basis": "announcement page", "weights_date": "2025-05-30", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenMMLab open reproduction of Grounding DINO for open-vocabulary detection and grounding.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2401.02361", "https://huggingface.co/openmmlab-community/mm_grounding_dino_large_all", "https://github.com/open-mmlab/mmdetection/tree/main/configs/mm_grounding_dino" ], "unverified": [] }, { "id": "glip", "name": "GLIP", "level": "model", "family_id": null, "developer": [ "microsoft-research", "ucla" ], "model_class": "vision", "modalities_in": [ "image", "text" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-12-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Grounded language-image pretraining that unifies detection and phrase grounding.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2112.03857", "https://github.com/open-mmlab/mmdetection/tree/main/configs/glip", "https://arxiv.org/html/2112.03857" ], "unverified": [] }, { "id": "meta-detic", "name": "Detic", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image", "text" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-01-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Detector trained with image-level labels to recognise twenty thousand classes, from the paper \"Detecting Twenty-thousand Classes using Image-level Supervision\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2201.02605", "https://github.com/open-mmlab/mmdetection/tree/main/projects/Detic" ], "unverified": [] }, { "id": "google-owl-vit", "name": "OWL-ViT", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image", "text" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "OWL-ViT B/32, B/16, L/14; OWLv2 base and large (ensemble and fine-tuned).", "context_length": null, "release_date": "2022-05-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google open-vocabulary object detectors built on CLIP-style vision transformers, prompted by text or an example image.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2205.06230", "https://github.com/google-research/scenic", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://huggingface.co/google/owlv2-base-patch16-ensemble", "https://huggingface.co/google/owlvit-base-patch32", "https://aihub.qualcomm.com/models/owl_vit", "https://github.com/google-research/scenic/tree/main/scenic/projects/owl_vit" ], "unverified": [] }, { "id": "google-owlvit-base-patch32", "name": "OWL-ViT B/32", "level": "variant", "family_id": "google-owl-vit", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image", "text" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": 153232471, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-05-12", "release_date_basis": "announcement page", "weights_date": "2022-07-05", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First OWL-ViT with a ViT-B/32 backbone.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2205.06230", "https://huggingface.co/google/owlvit-base-patch32" ], "unverified": [ "modalities_out" ] }, { "id": "google-owlv2-base-patch16-ensemble", "name": "OWLv2 B/16 ensemble", "level": "variant", "family_id": "google-owl-vit", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image", "text" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 154966791, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-06-16", "release_date_basis": "announcement page", "weights_date": "2023-10-13", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OWLv2 trained with self-training on web image-text pairs, from the paper \"Scaling Open-Vocabulary Object Detection\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2306.09683", "https://huggingface.co/google/owlv2-base-patch16-ensemble" ], "unverified": [] }, { "id": "google-owlv2-large-patch14", "name": "OWLv2 L/14", "level": "variant", "family_id": "google-owl-vit", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image", "text" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": 437610759, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-06-16", "release_date_basis": "announcement page", "weights_date": "2023-10-13", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Largest OWLv2 with a ViT-L/14 backbone.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2306.09683", "https://huggingface.co/google/owlv2-large-patch14" ], "unverified": [ "modalities_out" ] }, { "id": "google-f-vlm", "name": "F-VLM", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image", "text" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-09-30", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google open-vocabulary detector built on frozen vision-language models; deployable from Vertex AI Model Garden.", "verified_on": "2026-10-12", "sources": [ "https://github.com/google-research/google-research/tree/master/fvlm", "https://github.com/google-research/google-research", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://www.kaggle.com/models/googleai/f-vlm", "https://arxiv.org/abs/2209.15639" ], "unverified": [] }, { "id": "omdet-turbo", "name": "OmDet-Turbo", "level": "model", "family_id": null, "developer": [ "zhejiang-university" ], "model_class": "vision", "modalities_in": [ "image", "text" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 115428310, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-03-11", "release_date_basis": "announcement page", "weights_date": "2024-09-24", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time transformer open-vocabulary detector with an efficient fusion head.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2403.06892", "https://huggingface.co/omlab/omdet-turbo-swin-tiny-hf", "https://huggingface.co/api/models?pipeline_tag=zero-shot-object-detection&sort=downloads&limit=50", "https://arxiv.org/html/2403.06892" ], "unverified": [] }, { "id": "llmdet", "name": "LLMDet", "level": "model", "family_id": null, "developer": [ "sun-yat-sen-university", "alibaba-group" ], "model_class": "vision", "modalities_in": [ "image", "text" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 233473940, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-01-31", "release_date_basis": "announcement page", "weights_date": "2025-05-30", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open-vocabulary detector trained with supervision from a large language model, from Sun Yat-sen University.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2501.18954", "https://huggingface.co/iSEE-Laboratory/llmdet_base", "https://huggingface.co/api/models?pipeline_tag=zero-shot-object-detection&sort=downloads&limit=50", "https://arxiv.org/html/2501.18954" ], "unverified": [] }, { "id": "tencent-wedetect", "name": "WeDetect", "level": "model", "family_id": null, "developer": [ "tencent" ], "model_class": "vision", "modalities_in": [ "image", "text" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-12-16", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open-vocabulary detection models from the WeChat computer vision team at Tencent.", "verified_on": "2026-10-12", "sources": [ "https://github.com/WeChatCV/WeDetect", "https://aihub.qualcomm.com/models/wedetect" ], "unverified": [] }, { "id": "nvidia-locateanything-3b", "name": "LocateAnything 3B", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "vision", "modalities_in": [ "image", "text" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 3830665968, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-05-26", "release_date_basis": "announcement page", "weights_date": "2026-03-02", "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA vision-language grounding model that returns boxes for objects described in text, with parallel box decoding.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2605.27365", "https://huggingface.co/nvidia/LocateAnything-3B", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://huggingface.co/api/models?pipeline_tag=image-text-to-text&sort=likes&limit=50" ], "unverified": [] }, { "id": "recognize-anything", "name": "Recognize Anything (RAM)", "level": "model", "family_id": null, "developer": [ "oppo", "idea-research" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-06-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Image tagging model that recognises common categories, from the paper \"Recognize Anything: A Strong Image Tagging Model\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2306.03514", "https://replicate.com/idea-research/ram-grounded-sam", "https://arxiv.org/html/2306.03514" ], "unverified": [] }, { "id": "tii-falcon-perception", "name": "Falcon Perception", "level": "model", "family_id": null, "developer": [ "tii" ], "model_class": "vision", "modalities_in": [ "image", "text" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 632372288, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-03-28", "release_date_basis": "announcement page", "weights_date": "2026-02-22", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "TII open-vocabulary perception model for detection and segmentation from text prompts.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2603.27365", "https://huggingface.co/tiiuae/Falcon-Perception", "https://huggingface.co/api/models?pipeline_tag=mask-generation&sort=downloads&limit=50" ], "unverified": [] }, { "id": "fcn", "name": "FCN (Fully Convolutional Network)", "level": "model", "family_id": null, "developer": [ "uc-berkeley" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "FCN-32s, 16s, 8s; TorchVision FCN-ResNet50 and 101.", "context_length": null, "release_date": "2014-11-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fully convolutional network for semantic segmentation from UC Berkeley; the first end-to-end pixel labelling CNN and still a TorchVision baseline.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1411.4038", "https://github.com/shelhamer/fcn.berkeleyvision.org", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://aihub.qualcomm.com/models/fcn_resnet50", "https://github.com/pytorch/vision/blob/main/torchvision/models/segmentation/fcn.py", "https://pytorch.org/hub/pytorch_vision_fcn_resnet101", "https://raw.githubusercontent.com/pytorch/hub/master/pytorch_vision_fcn_resnet101.md", "https://arxiv.org/html/1411.4038" ], "unverified": [] }, { "id": "segnet", "name": "SegNet", "level": "model", "family_id": null, "developer": [ "university-of-cambridge" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2015-05-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Encoder-decoder network for semantic segmentation from the University of Cambridge, from the paper \"SegNet\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1511.00561", "https://github.com/alexgkendall/caffe-segnet", "https://modelzoo.co/model/segnet-caffe", "https://paperswithcode.com/method/segnet", "https://huggingface.co/datasets/pwc-archive/methods", "http://arxiv.org/abs/1511.00561v3", "https://arxiv.org/html/1511.00561", "https://arxiv.org/abs/1505.07293" ], "unverified": [ "developer" ] }, { "id": "u-net", "name": "U-Net", "level": "family", "family_id": null, "developer": [ "university-of-freiburg" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "U-Net 2D and 3D, UNet++ (arXiv 1807.10165), medical builds in NVIDIA DLE, OpenVINO BraTS model and TensorFlow volumetric models.", "context_length": null, "release_date": "2015-05-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU", "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Encoder-decoder segmentation network with a contracting path and a symmetric expanding path, first shown on biomedical images.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1505.04597", "https://aihub.qualcomm.com/models/unet_segmentation", "https://github.com/milesial/Pytorch-UNet", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/brain-tumor-segmentation-0002", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/dle/models/unet_2d-medical_tf1_ckpt", "https://pytorch.org/hub/mateuszbuda_brain-segmentation-pytorch_unet", "https://raw.githubusercontent.com/pytorch/hub/master/mateuszbuda_brain-segmentation-pytorch_unet.md", "https://arxiv.org/html/1505.04597" ], "unverified": [] }, { "id": "u-net-original", "name": "U-Net (2D)", "level": "variant", "family_id": "u-net", "developer": [ "university-of-freiburg" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2015-05-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "The original 2D U-Net, from the paper \"U-Net: Convolutional Networks for Biomedical Image Segmentation\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1505.04597", "https://github.com/milesial/Pytorch-UNet", "https://arxiv.org/html/1505.04597" ], "unverified": [] }, { "id": "unet-plus-plus", "name": "UNet++", "level": "variant", "family_id": "u-net", "developer": [ "arizona-state-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-07-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Nested U-Net with dense skip pathways for medical image segmentation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1807.10165", "https://modelzoo.co/model/pytorch-nested-unet", "https://github.com/4uiiurz1/pytorch-nested-unet", "https://paperswithcode.com/method/unet", "https://paperswithcode.com/method/unetr", "https://huggingface.co/datasets/pwc-archive/methods", "http://arxiv.org/abs/1807.10165v1", "https://arxiv.org/html/1807.10165" ], "unverified": [] }, { "id": "google-3d-unet", "name": "3D U-Net (TensorFlow volumetric models)", "level": "variant", "family_id": "u-net", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Volumetric U-Net for 3D medical image segmentation in the TensorFlow Model Garden.", "verified_on": "2026-10-12", "sources": [ "https://github.com/tensorflow/models/tree/master/official/projects/volumetric_models" ], "unverified": [] }, { "id": "pspnet", "name": "PSPNet", "level": "model", "family_id": null, "developer": [ "cuhk", "sensetime" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2016-12-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU", "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Semantic segmentation with a pyramid pooling module, from the paper \"Pyramid Scene Parsing Network\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1612.01105", "https://aihub.qualcomm.com/models/pspnet", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/pspnet-pytorch", "https://modelzoo.co/model/pspnet-pytorch", "https://github.com/Lextal/pspnet-pytorch", "https://paperswithcode.com/method/pspnet", "https://huggingface.co/datasets/pwc-archive/methods", "https://arxiv.org/html/1612.01105" ], "unverified": [] }, { "id": "google-deeplab", "name": "DeepLab", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "DeepLabV3, DeepLabV3+ (Xception, MobileNet, ResNet backbones), DeepLab-EdgeTPU and mobile 257x257 builds.", "context_length": null, "release_date": "2014-12-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU", "CPU server", "microcontroller" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, OpenVINO, Apple Core ML, STM32 model zoo, NXP eIQ model zoo, TI edge AI model zoo, MediaPipe.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google semantic segmentation family using atrous convolution and spatial pyramid pooling; DeepLabV3 with MobileNet runs on phones and microcontrollers.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1706.05587", "https://aihub.qualcomm.com/models/deeplab_xception", "https://github.com/LikeLy-Journey/SegmenTron", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/deeplabv3", "https://developer.apple.com/machine-learning/models/", "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/semantic_segmentation/deeplabv3", "https://github.com/NXP/eiq-model-zoo/tree/main/tasks/vision/semantic-segmentation", "https://arxiv.org/abs/1412.7062" ], "unverified": [] }, { "id": "google-deeplabv3", "name": "DeepLabV3", "level": "variant", "family_id": "google-deeplab", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-06-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "DeepLab with improved atrous spatial pyramid pooling, from the paper \"Rethinking Atrous Convolution for Semantic Image Segmentation\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1706.05587" ], "unverified": [] }, { "id": "google-deeplabv3-plus", "name": "DeepLabV3+", "level": "variant", "family_id": "google-deeplab", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-02-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "DeepLabV3 with a decoder, from the paper \"Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1802.02611" ], "unverified": [] }, { "id": "google-autoseg-edgetpu", "name": "AutoSeg-EdgeTPU", "level": "variant", "family_id": "google-deeplab", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Segmentation models searched for the Google Edge TPU, published on Kaggle with DeepLab-EdgeTPU.", "verified_on": "2026-10-12", "sources": [ "https://www.kaggle.com/models/google/autoseg-edgetpu", "https://github.com/tensorflow/models/tree/master/official/projects/edgetpu" ], "unverified": [] }, { "id": "lraspp", "name": "LR-ASPP", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-05-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Lite reduced atrous spatial pyramid pooling segmentation head introduced with MobileNet v3; TorchVision ships LR-ASPP MobileNetV3.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1905.02244", "https://pytorch.org/vision/stable/models/lraspp.html", "https://arxiv.org/html/1905.02244" ], "unverified": [] }, { "id": "icnet", "name": "ICNet", "level": "model", "family_id": null, "developer": [ "cuhk" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-04-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time semantic segmentation on high-resolution images with an image cascade.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1704.08545", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/icnet-camvid-ava-0001", "https://github.com/open-mmlab/mmsegmentation/tree/main/configs/icnet", "https://arxiv.org/html/1704.08545" ], "unverified": [] }, { "id": "bisenet", "name": "BiSeNet", "level": "model", "family_id": null, "developer": [ "hust", "megvii" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "BiSeNet V1 and V2 (arXiv 2004.02147); Qualcomm AI Hub phone build.", "context_length": null, "release_date": "2018-08-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Bilateral network for real-time semantic segmentation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1808.00897", "https://huggingface.co/qualcomm/BiseNet", "https://github.com/open-mmlab/mmsegmentation/tree/main/configs/bisenetv2", "https://arxiv.org/abs/2004.02147", "https://github.com/open-mmlab/mmsegmentation/tree/main/configs/bisenetv1", "https://arxiv.org/html/1808.00897" ], "unverified": [] }, { "id": "stdc", "name": "STDC", "level": "model", "family_id": null, "developer": [ "meituan" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-04-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time semantic segmentation network, from the paper \"Rethinking BiSeNet For Real-time Semantic Segmentation\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2104.13188", "https://github.com/open-mmlab/mmsegmentation/tree/main/configs/stdc", "https://arxiv.org/html/2104.13188" ], "unverified": [] }, { "id": "fast-scnn", "name": "Fast-SCNN", "level": "model", "family_id": null, "developer": [ "toshiba", "university-of-cambridge" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-02-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fast semantic segmentation network for embedded devices.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1902.04502", "https://github.com/open-mmlab/mmsegmentation/tree/main/configs/fastscnn", "https://arxiv.org/html/1902.04502" ], "unverified": [ "developer" ] }, { "id": "cgnet", "name": "CGNet", "level": "model", "family_id": null, "developer": [ "ict-cas" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-11-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Light-weight context guided network for semantic segmentation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1811.08201", "https://github.com/open-mmlab/mmsegmentation/tree/main/configs/cgnet", "https://arxiv.org/html/1811.08201" ], "unverified": [] }, { "id": "erfnet", "name": "ERFNet", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Efficient residual factorised network for real-time road scene segmentation (Romera et al.); in the OpenVINO Open Model Zoo.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/erfnet", "https://github.com/open-mmlab/mmsegmentation/tree/main/configs/erfnet" ], "unverified": [] }, { "id": "ddrnet", "name": "DDRNet", "level": "model", "family_id": null, "developer": [ "harbin-institute-of-technology" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-01-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Deep dual-resolution network for real-time road scene segmentation; DDRNet23-Slim runs on Qualcomm AI Hub.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2101.06085", "https://aihub.qualcomm.com/models/ddrnet23_slim", "https://github.com/chenjun2hao/DDRNet.pytorch", "https://github.com/open-mmlab/mmsegmentation/tree/main/configs/ddrnet", "https://arxiv.org/html/2101.06085" ], "unverified": [] }, { "id": "pidnet", "name": "PIDNet", "level": "model", "family_id": null, "developer": [ "texas-a-and-m-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-06-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time semantic segmentation network inspired by PID controllers.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2206.02066", "https://github.com/XuJiacong/PIDNet", "https://aihub.qualcomm.com/models/pidnet", "https://github.com/open-mmlab/mmsegmentation/tree/main/configs/pidnet", "https://arxiv.org/html/2206.02066" ], "unverified": [] }, { "id": "qualcomm-ffnet", "name": "FFNet", "level": "model", "family_id": null, "developer": [ "qualcomm" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-06-16", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qualcomm AI Research simple fuss-free networks for efficient semantic segmentation on devices.", "verified_on": "2026-10-12", "sources": [ "https://github.com/Qualcomm-AI-research/FFNet", "https://aihub.qualcomm.com/models/ffnet_40s", "https://arxiv.org/abs/2206.08236" ], "unverified": [] }, { "id": "baidu-pp-mobileseg", "name": "PP-MobileSeg", "level": "model", "family_id": null, "developer": [ "baidu" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "hybrid", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-04-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fast and accurate semantic segmentation model for mobile devices from the PaddleSeg team.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2304.05152", "https://github.com/open-mmlab/mmsegmentation/tree/main/projects/pp_mobileseg", "https://arxiv.org/html/2304.05152" ], "unverified": [] }, { "id": "fastseg", "name": "FastSeg", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MobileNetV3-based real-time semantic segmentation (large and small) in the OpenVINO Open Model Zoo.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/fastseg-large" ], "unverified": [] }, { "id": "duc", "name": "DUC (Dense Upsampling Convolution)", "level": "model", "family_id": null, "developer": [ "tusimple" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-02-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Semantic segmentation with dense upsampling and hybrid dilated convolution; in the ONNX Model Zoo.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1702.08502", "https://github.com/onnx/models/tree/main/validated/vision/object_detection_segmentation/duc", "https://arxiv.org/html/1702.08502" ], "unverified": [] }, { "id": "drn", "name": "DRN (Dilated Residual Networks)", "level": "model", "family_id": null, "developer": [ "princeton-university", "intel" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Dilated residual network DRN-D-38 for semantic segmentation, in the OpenVINO Open Model Zoo.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/drn-d-38", "https://arxiv.org/html/1705.09914" ], "unverified": [] }, { "id": "segformer", "name": "SegFormer", "level": "family", "family_id": null, "developer": [ "nvidia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "MiT-B0 to B5 on ADE20K and Cityscapes; NVIDIA TAO CitySemSegFormer and PeopleSemSegFormer.", "context_length": null, "release_date": "2021-05-31", "release_date_basis": "announcement page", "weights_date": "2022-03-02", "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, TI edge AI model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA simple and efficient transformer for semantic segmentation with a hierarchical Mix Transformer encoder and an MLP decoder.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2105.15203", "https://huggingface.co/nvidia/segformer-b0-finetuned-ade-512-512", "https://ai.azure.com/catalog/models?task=image-segmentation", "https://huggingface.co/nvidia/segformer-b5-finetuned-ade-640-640", "https://aihub.qualcomm.com/models/segformer_base", "https://github.com/huggingface/transformers/tree/main/src/transformers/models/segformer", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/citysemsegformer" ], "unverified": [] }, { "id": "nvidia-segformer-b0", "name": "SegFormer-B0", "level": "variant", "family_id": "segformer", "developer": [ "nvidia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 3800000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-05-31", "release_date_basis": "announcement page", "weights_date": "2022-03-02", "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Smallest SegFormer, fine-tuned on ADE20K at 512 pixels.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2105.15203", "https://huggingface.co/nvidia/segformer-b0-finetuned-ade-512-512" ], "unverified": [] }, { "id": "nvidia-segformer-b5", "name": "SegFormer-B5", "level": "variant", "family_id": "segformer", "developer": [ "nvidia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 85000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-05-31", "release_date_basis": "announcement page", "weights_date": "2022-03-02", "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Largest SegFormer, fine-tuned on ADE20K at 640 pixels.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2105.15203", "https://huggingface.co/nvidia/segformer-b5-finetuned-ade-640-640" ], "unverified": [] }, { "id": "segmenter", "name": "Segmenter", "level": "model", "family_id": null, "developer": [ "inria" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-05-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Pure transformer semantic segmentation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2105.05633", "https://github.com/open-mmlab/mmsegmentation/tree/main/configs/segmenter", "https://arxiv.org/html/2105.05633" ], "unverified": [ "developer" ] }, { "id": "setr", "name": "SETR", "level": "model", "family_id": null, "developer": [ "fudan-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-12-31", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Semantic segmentation from a sequence-to-sequence perspective with transformers.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2012.15840", "https://github.com/open-mmlab/mmsegmentation/tree/main/configs/setr", "https://arxiv.org/html/2012.15840" ], "unverified": [] }, { "id": "upernet", "name": "UPerNet", "level": "model", "family_id": null, "developer": [ "peking-university", "massachusetts-institute-of-technology" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": 60255436, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-07-26", "release_date_basis": "announcement page", "weights_date": "2023-01-13", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Unified perceptual parsing network; the standard segmentation head for evaluating backbones on ADE20K.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1807.10221", "https://huggingface.co/openmmlab/upernet-convnext-tiny", "https://ai.azure.com/catalog/models?task=image-segmentation", "https://github.com/open-mmlab/mmsegmentation/tree/main/configs/upernet", "https://arxiv.org/html/1807.10221" ], "unverified": [] }, { "id": "ocrnet", "name": "OCRNet", "level": "model", "family_id": null, "developer": [ "microsoft-research", "ict-cas" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-09-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Semantic segmentation with object-contextual representations.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1909.11065", "https://github.com/open-mmlab/mmsegmentation/tree/main/configs/ocrnet", "https://arxiv.org/html/1909.11065" ], "unverified": [] }, { "id": "segnext", "name": "SegNeXt", "level": "model", "family_id": null, "developer": [ "tsinghua-university", "nankai-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-09-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Semantic segmentation with convolutional attention.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2209.08575", "https://github.com/open-mmlab/mmsegmentation/tree/main/configs/segnext", "https://arxiv.org/html/2209.08575" ], "unverified": [] }, { "id": "attention-segmentation-networks", "name": "Attention-based segmentation networks (MMSegmentation)", "level": "model", "family_id": null, "developer": [ "casia", "jd-com" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "DANet (arXiv 1809.02983), CCNet (1811.11721), EMANet (1907.13426), ANN (1908.07678), APCNet, DMNet, DNLNet (2006.06668), EncNet (1803.08904), FastFCN (1903.11816), ISANet (1907.12273), NonLocal Net (1711.07971), PSANet.", "context_length": null, "release_date": "2018-09-09", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Group of semantic segmentation networks that add attention or context modules to a dilated ResNet; all are reproduced with checkpoints in OpenMMLab MMSegmentation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1809.02983", "https://github.com/open-mmlab/mmsegmentation/tree/main/configs/dnlnet", "https://arxiv.org/abs/2006.06668", "https://github.com/open-mmlab/mmsegmentation/tree/main/configs/isanet", "https://arxiv.org/abs/1907.12273", "https://github.com/open-mmlab/mmsegmentation/tree/main/configs/fastfcn", "https://arxiv.org/abs/1903.11816", "https://arxiv.org/html/1809.02983" ], "unverified": [] }, { "id": "meta-maskformer", "name": "MaskFormer", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 41783122, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-07-13", "release_date_basis": "announcement page", "weights_date": "2022-03-02", "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta mask classification model for semantic and panoptic segmentation, from the paper \"Per-Pixel Classification is Not All You Need\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2107.06278", "https://huggingface.co/facebook/maskformer-swin-tiny-coco", "https://ai.azure.com/catalog/models?task=image-segmentation", "https://github.com/open-mmlab/mmdetection/tree/main/configs/maskformer", "https://github.com/facebookresearch/MaskFormer" ], "unverified": [] }, { "id": "meta-mask2former", "name": "Mask2Former", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 216019734, "params_active": null, "params_checkpoint": null, "params_note": "Swin T to L for COCO, ADE20K, Cityscapes; video instance segmentation version (arXiv 2112.10764).", "context_length": null, "release_date": "2021-12-02", "release_date_basis": "announcement page", "weights_date": "2023-01-02", "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta universal segmentation model for semantic, instance and panoptic segmentation with masked attention.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2112.01527", "https://huggingface.co/facebook/mask2former-swin-large-coco-instance", "https://ai.azure.com/catalog/models?task=image-segmentation", "https://aihub.qualcomm.com/models/mask2former", "https://github.com/huggingface/transformers/tree/main/src/transformers/models/mask2former", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/mask2former", "https://huggingface.co/api/models?pipeline_tag=image-segmentation&sort=downloads&limit=50" ], "unverified": [] }, { "id": "oneformer", "name": "OneFormer", "level": "model", "family_id": null, "developer": [ "university-of-oregon", "uiuc" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-11-10", "release_date_basis": "announcement page", "weights_date": "2022-11-15", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "One transformer trained once for semantic, instance and panoptic segmentation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2211.06220", "https://huggingface.co/shi-labs/oneformer_ade20k_swin_large", "https://ai.azure.com/catalog/models?task=image-segmentation", "https://huggingface.co/shi-labs/oneformer_cityscapes_swin_large", "https://huggingface.co/api/models?pipeline_tag=image-segmentation&sort=downloads&limit=50", "https://arxiv.org/html/2211.06220" ], "unverified": [] }, { "id": "eomt", "name": "EoMT", "level": "model", "family_id": null, "developer": [ "eindhoven-university-of-technology" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 316695824, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03-24", "release_date_basis": "announcement page", "weights_date": "2025-03-26", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Encoder-only mask transformer that segments with a plain ViT, from the paper \"Your ViT is Secretly an Image Segmentation Model\" (TU Eindhoven).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2503.19108", "https://huggingface.co/tue-mps/coco_panoptic_eomt_large_640", "https://huggingface.co/api/models?pipeline_tag=image-segmentation&sort=downloads&limit=50", "https://arxiv.org/html/2503.19108" ], "unverified": [] }, { "id": "x-decoder", "name": "X-Decoder", "level": "model", "family_id": null, "developer": [ "university-of-wisconsin-madison", "ucla", "microsoft-research" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-12-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Generalised decoder for pixel, image and language tasks including open-vocabulary segmentation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2212.11270", "https://github.com/open-mmlab/mmdetection/tree/main/configs/XDecoder", "https://arxiv.org/html/2212.11270" ], "unverified": [] }, { "id": "san", "name": "SAN (Side Adapter Network)", "level": "model", "family_id": null, "developer": [ "hust", "microsoft-research" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-02-23", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open-vocabulary semantic segmentation with a side adapter on a frozen CLIP.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2302.12242", "https://github.com/open-mmlab/mmsegmentation/tree/main/configs/san", "https://arxiv.org/html/2302.12242" ], "unverified": [] }, { "id": "cat-seg", "name": "CAT-Seg", "level": "model", "family_id": null, "developer": [ "korea-university", "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-03-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open-vocabulary semantic segmentation by cost aggregation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2303.11797", "https://github.com/open-mmlab/mmsegmentation/tree/main/projects/CAT-Seg", "https://arxiv.org/html/2303.11797" ], "unverified": [ "developer" ] }, { "id": "vpd", "name": "VPD", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-03-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Visual perception (segmentation and depth) using a text-to-image diffusion model as backbone.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2303.02153", "https://github.com/open-mmlab/mmsegmentation/tree/main/configs/vpd" ], "unverified": [] }, { "id": "clipseg", "name": "CLIPSeg", "level": "model", "family_id": null, "developer": [ "university-of-gottingen" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 150748019, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-12-18", "release_date_basis": "announcement page", "weights_date": "2022-11-01", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Segmentation from text or image prompts on top of CLIP, from the paper \"Image Segmentation Using Text and Image Prompts\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2112.10003", "https://huggingface.co/CIDAS/clipseg-rd64-refined", "https://ai.azure.com/catalog/models?task=image-segmentation", "https://huggingface.co/api/models?pipeline_tag=image-segmentation&sort=downloads&limit=50", "https://arxiv.org/html/2112.10003" ], "unverified": [] }, { "id": "nvidia-odise", "name": "ODISE", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open-vocabulary panoptic segmentation with diffusion features from NVIDIA and UC San Diego, in the NVIDIA TAO catalog.", "verified_on": "2026-10-12", "sources": [ "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/odise" ], "unverified": [ "summary" ] }, { "id": "fastsam", "name": "FastSAM", "level": "model", "family_id": null, "developer": [ "casia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-06-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "agpl-3-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fast segment-anything model built on a YOLOv8 instance segmenter, from CASIA.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2306.12156", "https://github.com/CASIA-LMC-Lab/FastSAM", "https://aihub.qualcomm.com/models/fastsam_s", "https://docs.ultralytics.com/models/fast-sam/", "https://arxiv.org/html/2306.12156" ], "unverified": [] }, { "id": "mobilesam", "name": "MobileSAM", "level": "model", "family_id": null, "developer": [ "kyung-hee-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-06-25", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Lightweight Segment Anything for mobile applications, with a distilled small image encoder.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2306.14289", "https://aihub.qualcomm.com/models/mobilesam", "https://github.com/facebookresearch/segment-anything", "https://www.modelscope.cn/models/PAI/MobileSAM", "https://docs.ultralytics.com/models/mobile-sam/", "https://arxiv.org/html/2306.14289" ], "unverified": [] }, { "id": "sam-hq", "name": "SAM-HQ", "level": "model", "family_id": null, "developer": [ "eth-zurich", "hkust" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 94806832, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-06-02", "release_date_basis": "announcement page", "weights_date": "2025-02-18", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Segment Anything with higher-quality masks, from ETH Zurich and HKUST.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2306.01567", "https://huggingface.co/syscv-community/sam-hq-vit-base", "https://huggingface.co/api/models?pipeline_tag=mask-generation&sort=downloads&limit=50", "https://huggingface.co/yonigozlan/EdgeTAM-hf", "https://www.modelscope.cn/models/PAI/SAM-HQ", "https://arxiv.org/html/2306.01567" ], "unverified": [ "sources" ] }, { "id": "slimsam", "name": "SlimSAM", "level": "model", "family_id": null, "developer": [ "nus" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 9723784, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-12-08", "release_date_basis": "announcement page", "weights_date": "2024-01-07", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Pruned and distilled Segment Anything trained with 0.1 percent of the data.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2312.05284", "https://huggingface.co/nielsr/slimsam-77-uniform", "https://huggingface.co/api/models?pipeline_tag=mask-generation&sort=downloads&limit=50", "https://huggingface.co/yonigozlan/EdgeTAM-hf", "https://arxiv.org/html/2312.05284" ], "unverified": [ "sources" ] }, { "id": "meta-edgetam", "name": "EdgeTAM", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-01-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta on-device version of SAM 2 for tracking and segmenting objects in video on phones.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2501.07256", "https://github.com/facebookresearch/EdgeTAM", "https://aihub.qualcomm.com/models/edgetam" ], "unverified": [] }, { "id": "mit-han-lab-efficientvit-sam", "name": "EfficientViT-SAM", "level": "model", "family_id": null, "developer": [ "mit-han-lab" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-02-07", "release_date_basis": "announcement page", "weights_date": "2023-09-30", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MIT Han Lab Segment Anything with an EfficientViT image encoder for faster inference.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2402.05008", "https://huggingface.co/mit-han-lab/efficientvit-sam", "https://huggingface.co/api/models?pipeline_tag=mask-generation&sort=downloads&limit=50", "https://huggingface.co/yonigozlan/EdgeTAM-hf" ], "unverified": [] }, { "id": "track-anything", "name": "Track-Anything", "level": "model", "family_id": null, "developer": [ "sustech" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-04-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Interactive video object tracking and segmentation that combines Segment Anything with a video memory model.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2304.11968", "https://github.com/gaomingqi/Track-Anything", "https://aihub.qualcomm.com/models/track_anything", "https://arxiv.org/html/2304.11968" ], "unverified": [] }, { "id": "u2-net", "name": "U2-Net", "level": "model", "family_id": null, "developer": [ "university-of-alberta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-05-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Nested U-structure network for salient object detection, widely used for background removal.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2005.09007", "https://github.com/xuebinqin/U-2-Net", "https://aihub.qualcomm.com/models/u2net_segmentation", "https://www.modelscope.cn/models/iic/cv_u2net_salient-detection", "https://arxiv.org/html/2005.09007" ], "unverified": [] }, { "id": "basnet", "name": "BASNet", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Boundary-aware salient object detection network, published in KerasHub on Kaggle.", "verified_on": "2026-10-12", "sources": [ "https://www.kaggle.com/models/keras/basnet", "https://keras.io/keras_hub/presets/", "https://github.com/tensorflow/models/tree/master/official/projects/basnet" ], "unverified": [] }, { "id": "f3net", "name": "F3Net", "level": "model", "family_id": null, "developer": [ "ict-cas" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Salient object detection network with cascaded feedback decoder, in the OpenVINO Open Model Zoo.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/f3net", "https://arxiv.org/html/1911.11445" ], "unverified": [] }, { "id": "birefnet", "name": "BiRefNet", "level": "model", "family_id": null, "developer": [ "nankai-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 220700242, "params_active": null, "params_checkpoint": null, "params_note": "BiRefNet, lite, HR and HR-matting variants.", "context_length": null, "release_date": "2024-01-07", "release_date_basis": "announcement page", "weights_date": "2024-07-12", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "High-resolution dichotomous image segmentation with bilateral reference, widely used for background removal.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2401.03407", "https://huggingface.co/ZhengPeng7/BiRefNet", "https://huggingface.co/api/models?pipeline_tag=image-segmentation&sort=downloads&limit=50", "https://www.modelscope.cn/models/AI-ModelScope/BiRefNet", "https://arxiv.org/html/2401.03407" ], "unverified": [] }, { "id": "bria-rmbg", "name": "RMBG", "level": "model", "family_id": null, "developer": [ "bria-ai" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "other", "architecture_detail": null, "params_total": 44075590, "params_active": null, "params_checkpoint": null, "params_note": "RMBG 1.4 and 2.0 (2.0 is gated on Hugging Face).", "context_length": null, "release_date": "2023-12-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Background removal models from BRIA AI published on Hugging Face.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/briaai/RMBG-1.4", "https://huggingface.co/api/models?pipeline_tag=image-segmentation&sort=downloads&limit=50", "https://huggingface.co/briaai/RMBG-2.0", "https://www.modelscope.cn/models/AI-ModelScope/RMBG-2.0" ], "unverified": [] }, { "id": "prama-ben", "name": "BEN (Background Erase Network)", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "other", "architecture_detail": null, "params_total": 94632322, "params_active": null, "params_checkpoint": null, "params_note": "BEN and BEN2.", "context_length": null, "release_date": "2025-01-08", "release_date_basis": "announcement page", "weights_date": "2025-01-22", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Dichotomous image segmentation and matting model for background removal, from the paper \"BEN: Using Confidence-Guided Matting\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2501.06230", "https://huggingface.co/PramaLLC/BEN2", "https://huggingface.co/api/models?pipeline_tag=image-segmentation&sort=downloads&limit=50" ], "unverified": [] }, { "id": "deep-image-matting", "name": "DIM (Deep Image Matting)", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-03-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Image matting network that predicts alpha mattes from an image and trimap.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1703.03872", "https://github.com/open-mmlab/mmagic/tree/main/configs/dim" ], "unverified": [] }, { "id": "indexnet-matting", "name": "IndexNet", "level": "model", "family_id": null, "developer": [ "university-of-adelaide" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-08-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Image matting with learned index-guided pooling and upsampling.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1908.00672", "https://github.com/open-mmlab/mmagic/tree/main/configs/indexnet", "https://arxiv.org/html/1908.00672" ], "unverified": [] }, { "id": "gca-matting", "name": "GCA Matting", "level": "model", "family_id": null, "developer": [ "shanghai-jiao-tong-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-01-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Natural image matting with guided contextual attention.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2001.04069", "https://github.com/open-mmlab/mmagic/tree/main/configs/gca", "https://arxiv.org/html/2001.04069" ], "unverified": [] }, { "id": "robust-video-matting", "name": "Robust Video Matting", "level": "model", "family_id": null, "developer": [ "university-of-washington", "bytedance" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-08-25", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time high-resolution video matting with temporal guidance, used for background removal in video.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2108.11515", "https://replicate.com/arielreplicate/robust_video_matting", "https://arxiv.org/html/2108.11515" ], "unverified": [] }, { "id": "alibaba-damo-universal-matting", "name": "Universal matting (DAMO)", "level": "model", "family_id": null, "developer": [ "alibaba-damo" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba DAMO Academy U-Net matting model for general foreground extraction, published on ModelScope.", "verified_on": "2026-10-12", "sources": [ "https://www.modelscope.cn/models/iic/cv_unet_universal-matting" ], "unverified": [] }, { "id": "depth-anything", "name": "Depth Anything", "level": "family", "family_id": null, "developer": [ "hku", "bytedance" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Depth Anything V1, V2 (Small, Base, Large, Giant; metric indoor and outdoor) and Depth Anything 3 (multi-view).", "context_length": null, "release_date": "2024-01-19", "release_date_basis": "announcement page", "weights_date": "2024-01-22", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, Apple Core ML.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Monocular depth estimation models trained on large-scale unlabelled images, used in robotics, mapping and AR.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2401.10891", "https://huggingface.co/LiheYoung/depth-anything-small-hf", "https://aihub.qualcomm.com/models/depth_anything", "https://github.com/huggingface/transformers/tree/main/src/transformers/models/depth_anything", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/nvdepthanythingv2", "https://developer.apple.com/machine-learning/models/", "https://huggingface.co/apple/coreml-depth-anything-v2-small", "https://arxiv.org/html/2401.10891" ], "unverified": [] }, { "id": "depth-anything-v2-small", "name": "Depth Anything V2 Small", "level": "variant", "family_id": "depth-anything", "developer": [ "hku", "bytedance" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 24800000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-06-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Smallest Depth Anything V2 model, released under Apache 2.0.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2406.09414", "https://huggingface.co/depth-anything/Depth-Anything-V2-Small", "https://arxiv.org/html/2406.09414" ], "unverified": [] }, { "id": "depth-anything-v2-large", "name": "Depth Anything V2 Large", "level": "variant", "family_id": "depth-anything", "developer": [ "hku", "bytedance" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 335300000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-06-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Large Depth Anything V2 model, released for non-commercial use.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2406.09414", "https://huggingface.co/depth-anything/Depth-Anything-V2-Large", "https://arxiv.org/html/2406.09414" ], "unverified": [] }, { "id": "depth-anything-3", "name": "Depth Anything 3", "level": "variant", "family_id": "depth-anything", "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 135366599, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-11-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Depth Anything 3 recovers depth and camera geometry from any number of views.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2511.10647", "https://huggingface.co/depth-anything/DA3-BASE" ], "unverified": [] }, { "id": "intel-dpt", "name": "DPT (Dense Prediction Transformer)", "level": "model", "family_id": null, "developer": [ "intel" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 341850305, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-03-24", "release_date_basis": "announcement page", "weights_date": "2022-03-02", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Intel ViT-based dense prediction model for monocular depth estimation and segmentation; MiDaS v3 is built on it.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.13413", "https://huggingface.co/Intel/dpt-large", "https://ai.azure.com/catalog/models?task=image-segmentation", "https://paperswithcode.com/method/dense-prediction-transformer", "https://huggingface.co/datasets/pwc-archive/methods", "https://arxiv.org/abs/2103.13413v1", "https://github.com/open-mmlab/mmsegmentation/tree/main/configs/dpt" ], "unverified": [] }, { "id": "intel-zoedepth", "name": "ZoeDepth", "level": "model", "family_id": null, "developer": [ "intel" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 345073307, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-02-23", "release_date_basis": "announcement page", "weights_date": "2024-04-30", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Metric depth estimation that combines relative and metric depth training.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2302.12288", "https://huggingface.co/Intel/zoedepth-nyu-kitti", "https://huggingface.co/api/models?pipeline_tag=depth-estimation&sort=downloads&limit=50" ], "unverified": [] }, { "id": "apple-depth-pro", "name": "Depth Pro", "level": "model", "family_id": null, "developer": [ "apple" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-10-02", "release_date_basis": "announcement page", "weights_date": "2024-10-03", "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Apple zero-shot metric monocular depth model that produces sharp high-resolution depth maps in under a second.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2410.02073", "https://huggingface.co/apple/DepthPro", "https://huggingface.co/api/models?pipeline_tag=depth-estimation&sort=likes&limit=50" ], "unverified": [] }, { "id": "marigold", "name": "Marigold Depth", "level": "model", "family_id": null, "developer": [ "eth-zurich" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": 865922244, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-12-04", "release_date_basis": "announcement page", "weights_date": "2023-12-05", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Monocular depth estimation repurposed from a Stable Diffusion image generator, from ETH Zurich.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2312.02145", "https://huggingface.co/prs-eth/marigold-depth-v1-0", "https://huggingface.co/api/models?pipeline_tag=depth-estimation&sort=likes&limit=50", "https://www.modelscope.cn/models/Damo_XR_Lab/cv_marigold_monocular-depth-estimation", "https://arxiv.org/html/2312.02145" ], "unverified": [] }, { "id": "tencent-depthcrafter", "name": "DepthCrafter", "level": "model", "family_id": null, "developer": [ "tencent" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-09-03", "release_date_basis": "announcement page", "weights_date": "2024-09-14", "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tencent video depth estimation model that produces temporally consistent depth for long open-world videos.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2409.02095", "https://huggingface.co/tencent/DepthCrafter", "https://huggingface.co/api/models?pipeline_tag=depth-estimation&sort=likes&limit=50" ], "unverified": [] }, { "id": "adabins", "name": "AdaBins", "level": "model", "family_id": null, "developer": [ "kaust" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "hybrid", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-11-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Monocular depth estimation using adaptive depth bins.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2011.14141", "https://github.com/open-mmlab/mmsegmentation/tree/main/projects/Adabins", "https://arxiv.org/html/2011.14141" ], "unverified": [ "developer" ] }, { "id": "newcrfs", "name": "NeWCRFs", "level": "model", "family_id": null, "developer": [ "alibaba-group" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-03-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Monocular depth estimation with neural window fully-connected CRFs.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2203.01502", "https://www.modelscope.cn/models/iic/cv_newcrfs_image-depth-estimation_indoor", "https://arxiv.org/html/2203.01502" ], "unverified": [] }, { "id": "fastdepth", "name": "FastDepth", "level": "model", "family_id": null, "developer": [ "massachusetts-institute-of-technology" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "microcontroller", "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: STM32 model zoo, TI edge AI model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fast monocular depth estimation for embedded systems (MIT); STMicroelectronics ports it to STM32.", "verified_on": "2026-10-12", "sources": [ "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/depth_estimation/fastdepth", "https://github.com/TexasInstruments/edgeai-tensorlab/tree/main/edgeai-modelzoo/models/vision/depth_estimation/nyudepthv2", "https://arxiv.org/html/1903.03273" ], "unverified": [] }, { "id": "fcrn-depth", "name": "FCRN depth", "level": "model", "family_id": null, "developer": [ "tum" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fully convolutional residual network for indoor depth prediction (Laina et al.), in the OpenVINO Open Model Zoo.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/fcrn-dp-nyu-depth-v2-tf", "https://arxiv.org/html/1606.00373" ], "unverified": [] }, { "id": "vid2depth", "name": "vid2depth", "level": "model", "family_id": null, "developer": [ "ut-austin", "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-02-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Unsupervised learning of depth and ego-motion from monocular video.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1802.05522", "https://github.com/tensorflow/models/blob/master/research/README.md", "https://arxiv.org/html/1802.05522" ], "unverified": [] }, { "id": "google-ar-portrait-depth", "name": "AR Portrait Depth", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Portrait depth estimation model for augmented reality effects, published by TensorFlow on Kaggle.", "verified_on": "2026-10-12", "sources": [ "https://www.kaggle.com/models/tensorflow/ar-portrait-depth" ], "unverified": [] }, { "id": "crestereo", "name": "CREStereo", "level": "model", "family_id": null, "developer": [ "megvii" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Stereo matching network with cascaded recurrent refinement; a PyTorch port is distributed on GitHub.", "verified_on": "2026-10-12", "sources": [ "https://github.com/ibaiGorordo/CREStereo-Pytorch", "https://aihub.qualcomm.com/models/crestereo", "https://arxiv.org/html/2203.11483" ], "unverified": [] }, { "id": "stereonet", "name": "StereoNet", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time stereo depth network; a PyTorch implementation is distributed on GitHub.", "verified_on": "2026-10-12", "sources": [ "https://github.com/andrewlstewart/StereoNet_PyTorch", "https://aihub.qualcomm.com/models/stereonet", "https://arxiv.org/html/1807.08865" ], "unverified": [] }, { "id": "nvidia-foundationstereo", "name": "FoundationStereo", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "FoundationStereo and Fast-FoundationStereo.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA zero-shot stereo depth foundation model, published in the NGC TAO catalog.", "verified_on": "2026-10-12", "sources": [ "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/fast-foundationstereo" ], "unverified": [ "name" ] }, { "id": "nvidia-ess-bi3d", "name": "ESS and Bi3D (Isaac)", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "robot policy", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA Isaac ROS stereo disparity (ESS) and binary depth proximity segmentation (Bi3D) models for robot navigation.", "verified_on": "2026-10-12", "sources": [ "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/isaac/models/bi3d_proximity_segmentation" ], "unverified": [] }, { "id": "flownet", "name": "FlowNet", "level": "model", "family_id": null, "developer": [ "university-of-freiburg", "tum" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "FlowNet and FlowNet 2.0 (arXiv 1612.01925).", "context_length": null, "release_date": "2015-04-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First CNN for optical flow estimation, from the University of Freiburg.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1504.06852", "https://github.com/open-mmlab/mmflow/tree/master/configs/flownet", "https://github.com/open-mmlab/mmflow/tree/master/configs/flownet2", "https://arxiv.org/abs/1612.01925", "https://arxiv.org/html/1504.06852" ], "unverified": [] }, { "id": "liteflownet", "name": "LiteFlowNet", "level": "model", "family_id": null, "developer": [ "cuhk" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "LiteFlowNet and LiteFlowNet2 (arXiv 1903.07414).", "context_length": null, "release_date": "2018-05-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Lightweight CNN for optical flow estimation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1805.07036", "https://github.com/open-mmlab/mmflow/tree/master/configs/liteflownet", "https://github.com/open-mmlab/mmflow/tree/master/configs/liteflownet2", "https://arxiv.org/abs/1903.07414", "https://arxiv.org/html/1805.07036" ], "unverified": [] }, { "id": "pwc-net", "name": "PWC-Net", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-09-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Optical flow CNN using pyramid, warping and cost volume.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1709.02371", "https://github.com/open-mmlab/mmflow/tree/master/configs/pwcnet" ], "unverified": [] }, { "id": "raft", "name": "RAFT", "level": "model", "family_id": null, "developer": [ "princeton-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-03-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Optical flow with recurrent all-pairs field transforms, from Princeton University; TorchVision ships it.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2003.12039", "https://www.modelscope.cn/models/iic/cv_raft_video-frame-interpolation", "https://github.com/open-mmlab/mmflow/tree/master/configs/raft", "https://pytorch.org/vision/stable/models/raft.html", "https://arxiv.org/html/2003.12039" ], "unverified": [] }, { "id": "gma", "name": "GMA", "level": "model", "family_id": null, "developer": [ "australian-national-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-04-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Optical flow with global motion aggregation for occluded regions.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2104.02409", "https://github.com/open-mmlab/mmflow/tree/master/configs/gma", "https://arxiv.org/html/2104.02409" ], "unverified": [] }, { "id": "irr", "name": "IRR", "level": "model", "family_id": null, "developer": [ "tu-darmstadt" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-04-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Iterative residual refinement for joint optical flow and occlusion estimation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1904.05290", "https://github.com/open-mmlab/mmflow/tree/master/configs/irr", "https://arxiv.org/html/1904.05290" ], "unverified": [] }, { "id": "maskflownet", "name": "MaskFlownet", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-03-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Optical flow with asymmetric feature matching and a learnable occlusion mask.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2003.10955", "https://github.com/open-mmlab/mmflow/tree/master/configs/maskflownet" ], "unverified": [] }, { "id": "superpoint", "name": "SuperPoint", "level": "model", "family_id": null, "developer": [ "magic-leap" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": 1300865, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-12-20", "release_date_basis": "announcement page", "weights_date": "2024-03-13", "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Self-supervised interest point detector and descriptor from Magic Leap, used for visual localisation and 3D reconstruction.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1712.07629", "https://huggingface.co/magic-leap-community/superpoint", "https://aihub.qualcomm.com/models/superpoint", "https://github.com/rpautrat/SuperPoint", "https://huggingface.co/api/models?pipeline_tag=keypoint-detection&sort=downloads&limit=50", "https://huggingface.co/ETH-CVG/lightglue_superpoint", "https://arxiv.org/html/1712.07629" ], "unverified": [ "modalities_out" ] }, { "id": "lightglue", "name": "LightGlue", "level": "model", "family_id": null, "developer": [ "eth-zurich" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": 13744594, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-06-23", "release_date_basis": "announcement page", "weights_date": "2025-02-20", "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fast local feature matcher from ETH Zurich, used with SuperPoint or ALIKED keypoints for mapping and inspection.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2306.13643", "https://huggingface.co/ETH-CVG/lightglue_superpoint", "https://huggingface.co/api/models?pipeline_tag=keypoint-detection&sort=downloads&limit=50", "https://www.kaggle.com/models/oldufo/lightglue", "https://www.kaggle.com/models/oldufo/aliked" ], "unverified": [ "modalities_out" ] }, { "id": "loftr", "name": "LoFTR", "level": "model", "family_id": null, "developer": [ "zhejiang-university", "sensetime" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-04-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Detector-free local feature matching with transformers, from Zhejiang University.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2104.00680", "https://www.kaggle.com/models/jiamingsuen/loftr", "https://arxiv.org/html/2104.00680" ], "unverified": [ "modalities_out" ] }, { "id": "aliked", "name": "ALIKED", "level": "model", "family_id": null, "developer": [ "beihang-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-04-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Light keypoint and descriptor extraction network with deformable transformation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2304.03608", "https://www.kaggle.com/models/oldufo/aliked", "https://www.kaggle.com/models/oldufo/lightglue", "https://arxiv.org/html/2304.03608" ], "unverified": [] }, { "id": "hf-net", "name": "HF-Net", "level": "model", "family_id": null, "developer": [ "eth-zurich" ], "model_class": "robot policy", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-12-09", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Hierarchical localisation network from ETH Zurich that predicts global and local features for large-scale visual localisation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1812.03506", "https://github.com/ethz-asl/hfnet", "https://aihub.qualcomm.com/models/hfnet", "https://arxiv.org/html/1812.03506" ], "unverified": [] }, { "id": "netvlad", "name": "NetVLAD", "level": "model", "family_id": null, "developer": [ "inria" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Place recognition CNN with a trainable VLAD layer (Arandjelovic et al.), in the OpenVINO Open Model Zoo.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/netvlad-tf", "https://arxiv.org/html/1511.07247" ], "unverified": [ "developer" ] }, { "id": "megaloc", "name": "MegaLoc", "level": "model", "family_id": null, "developer": [ "polytechnic-of-turin" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 228639552, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-02-24", "release_date_basis": "announcement page", "weights_date": "2025-02-27", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Image retrieval model for visual place recognition and localisation across many settings.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2502.17237", "https://huggingface.co/gberton/MegaLoc", "https://huggingface.co/api/models?pipeline_tag=image-feature-extraction&sort=downloads&limit=50", "https://arxiv.org/html/2502.17237" ], "unverified": [] }, { "id": "google-delf", "name": "DELF (DEep Local Features)", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2016-12-19", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google attentive local features for large-scale image retrieval and landmark recognition.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1612.06321", "https://www.kaggle.com/models/google/delf", "https://github.com/tensorflow/models/blob/master/research/README.md" ], "unverified": [ "modalities_out" ] }, { "id": "google-planet-v2", "name": "PlaNet v2", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2016-02-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google photo geolocation classifier that predicts where on Earth a photo was taken.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1602.05314", "https://www.kaggle.com/models/google/planet-v2" ], "unverified": [] }, { "id": "openpose", "name": "OpenPose", "level": "model", "family_id": null, "developer": [ "carnegie-mellon-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2016-11-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Carnegie Mellon real-time multi-person 2D pose estimation of body, foot, hand and face keypoints using part affinity fields.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1812.08008", "https://github.com/CMU-Perceptual-Computing-Lab/openpose", "https://modelzoo.co/model/openpose-caffe", "https://modelzoo.co/model/pytorch-openpose", "https://paperswithcode.com/method/openpose", "https://huggingface.co/datasets/pwc-archive/methods", "https://arxiv.org/abs/1812.08008v2", "https://arxiv.org/abs/1611.08050" ], "unverified": [] }, { "id": "hrnet", "name": "HRNet", "level": "family", "family_id": null, "developer": [ "university-of-science-and-technology-of-china", "microsoft-research" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "HRNet-W32 and W48 for pose; HRNetV2 for classification and segmentation (arXiv 1908.07919); HigherHRNet; Lite-HRNet; HRFormer.", "context_length": null, "release_date": "2019-02-25", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU", "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "High-resolution networks that keep high-resolution features through the whole network, used for human pose estimation, segmentation and as backbones.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1902.09212", "https://github.com/leoxiaobin/deep-high-resolution-net.pytorch", "https://aihub.qualcomm.com/models/hrnet_pose", "https://aihub.qualcomm.com/models/litehrnet", "https://github.com/HRNet/Lite-HRNet", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/higher-hrnet-w32-human-pose-estimation", "https://www.modelscope.cn/models/iic/cv_hrnetv2w32_body-2d-keypoints_image", "https://arxiv.org/html/1902.09212" ], "unverified": [ "modalities_out" ] }, { "id": "hrnet-pose", "name": "HRNet (pose)", "level": "variant", "family_id": "hrnet", "developer": [ "university-of-science-and-technology-of-china", "microsoft-research" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-02-25", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Top-down human pose estimator, from the paper \"Deep High-Resolution Representation Learning for Human Pose Estimation\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1902.09212", "https://arxiv.org/html/1902.09212" ], "unverified": [ "modalities_out" ] }, { "id": "hrnetv2", "name": "HRNetV2", "level": "variant", "family_id": "hrnet", "developer": [ "microsoft-research" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-08-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "HRNet for classification, segmentation and detection, from the paper \"Deep High-Resolution Representation Learning for Visual Recognition\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1908.07919", "https://arxiv.org/html/1908.07919" ], "unverified": [] }, { "id": "higherhrnet", "name": "HigherHRNet", "level": "variant", "family_id": "hrnet", "developer": [ "uiuc", "microsoft-corporation" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-08-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Bottom-up multi-person pose estimation with scale-aware high-resolution features.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1908.10357", "https://arxiv.org/html/1908.10357" ], "unverified": [ "modalities_out" ] }, { "id": "lite-hrnet", "name": "Lite-HRNet", "level": "variant", "family_id": "hrnet", "developer": [ "hust", "microsoft-research" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-04-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Lightweight high-resolution network for pose estimation on mobile devices.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2104.06403", "https://github.com/HRNet/Lite-HRNet", "https://arxiv.org/html/2104.06403" ], "unverified": [] }, { "id": "hrformer", "name": "HRFormer", "level": "variant", "family_id": "hrnet", "developer": [ "peking-university", "microsoft-research" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-10-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "High-resolution transformer for dense prediction including pose estimation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2110.09408", "https://arxiv.org/html/2110.09408" ], "unverified": [] }, { "id": "simplebaseline", "name": "SimpleBaseline (pose)", "level": "model", "family_id": null, "developer": [ "microsoft-research" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "SimpleBaseline2D; SimpleBaseline3D is a different paper (arXiv 1705.03098) for 3D lifting.", "context_length": null, "release_date": "2018-04-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Simple deconvolution-head baseline for human pose estimation and tracking.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1804.06208", "https://github.com/open-mmlab/mmpose/tree/main/configs", "https://arxiv.org/abs/1705.03098", "https://arxiv.org/html/1804.06208" ], "unverified": [] }, { "id": "stacked-hourglass", "name": "Stacked Hourglass", "level": "model", "family_id": null, "developer": [ "university-of-michigan" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2016-03-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Repeated bottom-up, top-down network for human pose estimation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1603.06937", "https://github.com/open-mmlab/mmpose/tree/main/configs", "https://arxiv.org/html/1603.06937" ], "unverified": [] }, { "id": "cpm", "name": "CPM (Convolutional Pose Machines)", "level": "model", "family_id": null, "developer": [ "carnegie-mellon-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2016-01-30", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Sequential convolutional pose estimator that refines belief maps over stages.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1602.00134", "https://github.com/open-mmlab/mmpose/tree/main/configs", "https://arxiv.org/html/1602.00134" ], "unverified": [ "developer" ] }, { "id": "deeppose", "name": "DeepPose", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2013-12-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First deep network for human pose estimation by direct keypoint regression.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1312.4659", "https://github.com/open-mmlab/mmpose/tree/main/configs", "https://arxiv.org/html/1312.4659" ], "unverified": [] }, { "id": "associative-embedding", "name": "Associative Embedding", "level": "model", "family_id": null, "developer": [ "university-of-michigan", "tsinghua-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2016-11-16", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Bottom-up multi-person pose estimation that groups keypoints with learned tags.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1611.05424", "https://github.com/open-mmlab/mmpose/tree/main/configs", "https://arxiv.org/html/1611.05424" ], "unverified": [] }, { "id": "dekr", "name": "DEKR", "level": "model", "family_id": null, "developer": [ "university-of-science-and-technology-of-china", "microsoft-research" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-04-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Bottom-up pose estimation with disentangled keypoint regression.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2104.02300", "https://github.com/open-mmlab/mmpose/tree/main/configs", "https://arxiv.org/html/2104.02300" ], "unverified": [] }, { "id": "mspn", "name": "MSPN", "level": "model", "family_id": null, "developer": [ "megvii" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-01-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multi-stage pose network, from the paper \"Rethinking on Multi-Stage Networks for Human Pose Estimation\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1901.00148", "https://github.com/open-mmlab/mmpose/tree/main/configs", "https://arxiv.org/html/1901.00148" ], "unverified": [] }, { "id": "rsn", "name": "RSN", "level": "model", "family_id": null, "developer": [ "tsinghua-university", "megvii" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-03-09", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Residual steps network for delicate local representations in multi-person pose estimation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2003.04030", "https://github.com/open-mmlab/mmpose/tree/main/configs", "https://arxiv.org/html/2003.04030" ], "unverified": [ "modalities_out" ] }, { "id": "simcc", "name": "SimCC", "level": "model", "family_id": null, "developer": [ "tsinghua-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-07-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Pose estimation as coordinate classification along x and y.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2107.03332", "https://github.com/open-mmlab/mmpose/tree/main/configs", "https://arxiv.org/html/2107.03332" ], "unverified": [] }, { "id": "vipnas", "name": "ViPNAS", "level": "model", "family_id": null, "developer": [ "cuhk", "sensetime" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-05-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Efficient video pose estimation network found by architecture search.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2105.10154", "https://github.com/open-mmlab/mmpose/tree/main/configs", "https://arxiv.org/html/2105.10154" ], "unverified": [] }, { "id": "ed-pose", "name": "ED-Pose", "level": "model", "family_id": null, "developer": [ "idea-research" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-02-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "End-to-end multi-person pose estimation with explicit box detection.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2302.01593", "https://github.com/open-mmlab/mmpose/tree/main/configs", "https://arxiv.org/html/2302.01593" ], "unverified": [] }, { "id": "poseanything", "name": "PoseAnything", "level": "model", "family_id": null, "developer": [ "tel-aviv-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-11-29", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Category-agnostic pose estimation with a graph-based approach.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2311.17891", "https://github.com/open-mmlab/mmpose/tree/main/projects/pose_anything", "https://arxiv.org/html/2311.17891" ], "unverified": [] }, { "id": "interhand", "name": "InterNet (InterHand2.6M)", "level": "model", "family_id": null, "developer": [ "seoul-national-university", "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-08-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Baseline network for 3D interacting hand pose estimation from a single RGB image.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2008.09309", "https://github.com/open-mmlab/mmpose/tree/main/configs", "https://arxiv.org/html/2008.09309" ], "unverified": [] }, { "id": "videopose3d", "name": "VideoPose3D", "level": "model", "family_id": null, "developer": [ "eth-zurich", "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-11-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "3D human pose estimation in video with temporal convolutions.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1811.11742", "https://github.com/open-mmlab/mmpose/tree/main/configs", "https://arxiv.org/html/1811.11742" ], "unverified": [] }, { "id": "voxelpose", "name": "VoxelPose", "level": "model", "family_id": null, "developer": [ "microsoft-research", "university-of-science-and-technology-of-china" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-04-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multi-camera 3D human pose estimation in the wild.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2004.06239", "https://github.com/open-mmlab/mmpose/tree/main/configs", "https://arxiv.org/html/2004.06239" ], "unverified": [] }, { "id": "openmmlab-rtmpose", "name": "RTMPose", "level": "model", "family_id": null, "developer": [ "openmmlab" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "RTMPose T to L; RTMO (arXiv 2312.07526) one-stage; RTMW and RTMW3D whole-body (arXiv 2407.08634).", "context_length": null, "release_date": "2023-03-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenMMLab real-time multi-person pose estimation models for body, hand, face and whole-body keypoints.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2303.07399", "https://github.com/open-mmlab/mmpose", "https://aihub.qualcomm.com/models/rtmpose_body2d", "https://github.com/open-mmlab/mmpose/tree/main/projects/rtmpose" ], "unverified": [] }, { "id": "openmmlab-rtmo", "name": "RTMO", "level": "model", "family_id": null, "developer": [ "openmmlab" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-12-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenMMLab one-stage real-time multi-person pose estimator.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2312.07526", "https://github.com/open-mmlab/mmpose/tree/main/projects/rtmo" ], "unverified": [] }, { "id": "openmmlab-rtmw", "name": "RTMW", "level": "model", "family_id": null, "developer": [ "openmmlab" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-07-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenMMLab real-time multi-person 2D and 3D whole-body pose estimation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2407.08634", "https://github.com/open-mmlab/mmpose/tree/main/projects/rtmpose3d" ], "unverified": [] }, { "id": "openmmlab-yolox-pose", "name": "YOLOX-Pose", "level": "model", "family_id": null, "developer": [ "openmmlab" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "YOLOX detector extended with keypoints for one-stage pose estimation, an MMPose project.", "verified_on": "2026-10-12", "sources": [ "https://github.com/open-mmlab/mmpose/tree/main/projects/yolox_pose" ], "unverified": [] }, { "id": "vitpose", "name": "ViTPose", "level": "model", "family_id": null, "developer": [ "university-of-sydney", "jd-com" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 89995537, "params_active": null, "params_checkpoint": null, "params_note": "ViTPose B and ViTPose++ S, B, L, H.", "context_length": null, "release_date": "2022-04-26", "release_date_basis": "announcement page", "weights_date": "2025-01-08", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Plain vision transformer baselines for human pose estimation, from the University of Sydney.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2204.12484", "https://huggingface.co/usyd-community/vitpose-base", "https://huggingface.co/api/models?pipeline_tag=keypoint-detection&sort=downloads&limit=50", "https://huggingface.co/usyd-community/vitpose-plus-base", "https://github.com/open-mmlab/mmpose/tree/main/configs", "https://arxiv.org/html/2204.12484" ], "unverified": [] }, { "id": "alphapose", "name": "AlphaPose", "level": "model", "family_id": null, "developer": [ "shanghai-jiao-tong-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-01-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time whole-body multi-person pose estimation and tracking from Shanghai Jiao Tong University.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2211.03375", "https://github.com/MVIG-SJTU/AlphaPose", "https://modelzoo.co/model/alphapose", "https://arxiv.org/html/2211.03375" ], "unverified": [] }, { "id": "meta-sapiens", "name": "Sapiens", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Sapiens and Sapiens2 (0.4B, 1B, 5B; pretrain, pose, segmentation, depth, normals).", "context_length": null, "release_date": "2024-08-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta foundation models for human-centric vision: pose, body-part segmentation, depth and surface normals.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2408.12569", "https://huggingface.co/api/models?pipeline_tag=keypoint-detection&sort=downloads&limit=50", "https://huggingface.co/facebook/sapiens2-pretrain-1b", "https://www.modelscope.cn/models/facebook/sapiens2-seg-0.4b" ], "unverified": [ "modalities_out" ] }, { "id": "meta-sapiens2", "name": "Sapiens2", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": 1455474192, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04-23", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Second generation of Meta human-centric vision foundation models.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2604.21681", "https://huggingface.co/facebook/sapiens2-pretrain-1b", "https://huggingface.co/api/models?pipeline_tag=keypoint-detection&sort=downloads&limit=50" ], "unverified": [ "modalities_out" ] }, { "id": "google-movenet", "name": "MoveNet", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Lightning, Thunder and MultiPose; NXP and STM32 ports.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU", "microcontroller" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, STM32 model zoo, NXP eIQ model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google ultra-fast 17-keypoint pose estimation model for fitness and health apps on phones and edge devices.", "verified_on": "2026-10-12", "sources": [ "https://www.kaggle.com/models/google/movenet", "https://aihub.qualcomm.com/models/movenet", "https://github.com/lee-man/movenet-pytorch", "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet", "https://github.com/NXP/eiq-model-zoo/tree/main/tasks/vision/pose-estimation", "https://github.com/tensorflow/tfhub.dev/tree/master/assets/docs/google" ], "unverified": [] }, { "id": "google-posenet", "name": "PoseNet", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier TensorFlow.js pose estimation model with MobileNet and ResNet-50 backbones, published on Kaggle.", "verified_on": "2026-10-12", "sources": [ "https://www.kaggle.com/models/tensorflow/posenet-mobilenet", "https://aihub.qualcomm.com/models/posenet_mobilenet", "https://github.com/rwightman/posenet-pytorch", "https://www.kaggle.com/models/tensorflow/posenet-resnet50", "https://github.com/tensorflow/tfhub.dev/tree/master/assets/docs/tensorflow" ], "unverified": [ "modalities_out" ] }, { "id": "google-bodypix", "name": "BodyPix", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "TensorFlow.js person and body-part segmentation model.", "verified_on": "2026-10-12", "sources": [ "https://github.com/tensorflow/tfhub.dev/tree/master/assets/docs/tensorflow", "https://www.kaggle.com/models/tensorflow/bodypix" ], "unverified": [] }, { "id": "google-mediapipe", "name": "MediaPipe vision models", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "BlazeFace, BlazePose GHUM, Face Mesh V2 with blendshapes, Hands, Iris, gesture recogniser, selfie, hair and multiclass segmenters, interactive segmenter, holistic landmarker, object detector.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "mobile or NPU", "microcontroller" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, MediaPipe, STM32 model zoo, NXP eIQ model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google on-device vision models behind the MediaPipe solutions for faces, hands, pose, gestures and segmentation; run on phones, browsers and microcontrollers.", "verified_on": "2026-10-12", "sources": [ "https://ai.google.dev/edge/mediapipe/solutions/guide", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://aihub.qualcomm.com/models/mediapipe_face", "https://github.com/zmurez/MediaPipePyTorch/", "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/face_detection/facedetect_front", "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation", "https://github.com/NXP/eiq-model-zoo/tree/main/tasks/vision/semantic-segmentation" ], "unverified": [] }, { "id": "google-blazeface", "name": "BlazeFace", "level": "variant", "family_id": "google-mediapipe", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Short-range and full-range (dense and sparse).", "context_length": null, "release_date": "2019-07-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "microcontroller", "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: STM32 model zoo, MediaPipe.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Sub-millisecond face detector for mobile GPUs.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1907.05047", "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/face_detection/facedetect_front", "https://ai.google.dev/edge/mediapipe/solutions/vision/face_detector", "https://www.kaggle.com/models/tensorflow/blazeface" ], "unverified": [] }, { "id": "google-blazepose", "name": "BlazePose GHUM", "level": "variant", "family_id": "google-mediapipe", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Lite, Full and Heavy; 33 landmarks.", "context_length": null, "release_date": "2020-06-17", "release_date_basis": "announcement page", "weights_date": "2024-02-25", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: MediaPipe.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "On-device body pose tracking with 33 landmarks, from the paper \"BlazePose\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2006.10204", "https://huggingface.co/qualcomm/MediaPipe-Pose-Estimation", "https://huggingface.co/api/models?pipeline_tag=keypoint-detection&sort=downloads&limit=50", "https://ai.google.dev/edge/mediapipe/solutions/vision/pose_landmarker" ], "unverified": [] }, { "id": "google-mediapipe-face-mesh", "name": "Face Mesh", "level": "variant", "family_id": "google-mediapipe", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Face Mesh V2 (478 landmarks) with Blendshape V2; Attention Mesh (arXiv 2006.10962).", "context_length": null, "release_date": "2019-07-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: MediaPipe.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time 3D face surface landmark model for mobile GPUs.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1907.06724", "https://ai.google.dev/edge/mediapipe/solutions/vision/face_landmarker", "https://github.com/google-ai-edge/mediapipe/blob/master/docs/solutions/models.md", "https://arxiv.org/abs/2006.10962" ], "unverified": [] }, { "id": "google-mediapipe-hands", "name": "MediaPipe Hands", "level": "variant", "family_id": "google-mediapipe", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-06-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: MediaPipe.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "On-device palm detection and 21-point hand landmark tracking; the gesture recogniser builds on it.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2006.10214", "https://ai.google.dev/edge/mediapipe/solutions/vision/hand_landmarker", "https://ai.google.dev/edge/mediapipe/solutions/vision/gesture_recognizer", "https://www.kaggle.com/models/tensorflow/handskeleton" ], "unverified": [] }, { "id": "google-mediapipe-selfie-segmenter", "name": "MediaPipe Selfie Segmenter", "level": "variant", "family_id": "google-mediapipe", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Square and landscape selfie segmenters, multiclass (hair, skin, clothes) and hair segmenter.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: NXP eIQ model zoo, MediaPipe.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Person and selfie segmentation models for video calls and camera effects.", "verified_on": "2026-10-12", "sources": [ "https://ai.google.dev/edge/mediapipe/solutions/vision/image_segmenter", "https://github.com/NXP/eiq-model-zoo/tree/main/tasks/vision/semantic-segmentation" ], "unverified": [] }, { "id": "google-mediapipe-iris", "name": "MediaPipe Iris", "level": "variant", "family_id": "google-mediapipe", "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-06-19", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: MediaPipe.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time pupil and iris tracking from monocular video.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2006.11341", "https://github.com/google-ai-edge/mediapipe/blob/master/docs/solutions/models.md" ], "unverified": [] }, { "id": "insightface-arcface", "name": "ArcFace", "level": "model", "family_id": null, "developer": [ "insightface" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "ArcFace ResNet-50 and ResNet-100 (OpenVINO, ONNX, ModelScope).", "context_length": null, "release_date": "2018-01-23", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Face recognition model trained with additive angular margin loss, from the InsightFace project; the reference open face recogniser.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1801.07698", "https://github.com/deepinsight/insightface", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/face-recognition-resnet100-arcface-onnx", "https://www.modelscope.cn/models/iic/cv_ir50_face-recognition_arcface", "https://modelzoo.co/model/insightface", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/arcface", "https://github.com/onnx/models/tree/main/validated/vision/body_analysis/arcface" ], "unverified": [] }, { "id": "insightface-retinaface", "name": "RetinaFace", "level": "model", "family_id": null, "developer": [ "insightface" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-05-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Single-stage dense face localisation with landmarks, from the InsightFace project.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1905.00641", "https://github.com/deepinsight/insightface", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/retinaface-resnet50-pytorch", "https://www.modelscope.cn/models/iic/cv_resnet50_face-detection_retinaface" ], "unverified": [] }, { "id": "insightface-scrfd", "name": "SCRFD", "level": "model", "family_id": null, "developer": [ "insightface" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-05-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Efficient face detector from the InsightFace project, from the paper \"Sample and Computation Redistribution for Efficient Face Detection\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2105.04714", "https://github.com/deepinsight/insightface", "https://www.modelscope.cn/models/iic/cv_resnet_facedetection_scrfd10gkps" ], "unverified": [] }, { "id": "mtcnn", "name": "MTCNN", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2016-04-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Cascaded CNNs for joint face detection and alignment.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1604.02878", "https://www.modelscope.cn/models/iic/cv_manual_face-detection_mtcnn" ], "unverified": [] }, { "id": "facenet", "name": "FaceNet", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Inception-ResNet FaceNet (Sandberg implementation, OpenVINO) and FaceNet512 (deepface, NXP).", "context_length": null, "release_date": "2015-03-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server", "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO, NXP eIQ model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Face recognition and clustering with a learned 128-dimension embedding, from the paper \"FaceNet\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1503.03832", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/facenet-20180408-102900", "https://github.com/NXP/eiq-model-zoo/tree/main/tasks/vision/face-recognition", "https://modelzoo.co/model/facenet-pytorch", "https://github.com/timesler/facenet-pytorch", "https://arxiv.org/html/1503.03832" ], "unverified": [ "modalities_out" ] }, { "id": "mobilefacenet", "name": "MobileFaceNet", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-04-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Efficient CNN for real-time face verification on mobile devices.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1804.07573", "https://github.com/foamliu/MobileFaceNet", "https://aihub.qualcomm.com/models/mobile_facenet" ], "unverified": [] }, { "id": "ultraface", "name": "Ultra-Light-Fast-Generic-Face-Detector (UltraFace)", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "RFB-320, slim and ultraslim.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server", "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO, NXP eIQ model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Face detector of about 1MB for edge devices.", "verified_on": "2026-10-12", "sources": [ "https://github.com/Linzaer/Ultra-Light-Fast-Generic-Face-Detector-1MB", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/ultra-lightweight-face-detection-rfb-320", "https://github.com/NXP/eiq-model-zoo/tree/main/tasks/vision/object-detection", "https://github.com/onnx/models/tree/main/validated/vision/body_analysis/ultraface" ], "unverified": [] }, { "id": "faceboxes", "name": "FaceBoxes", "level": "model", "family_id": null, "developer": [ "casia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "CPU real-time face detector, in the OpenVINO Open Model Zoo.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/faceboxes-pytorch", "https://arxiv.org/html/1708.05234" ], "unverified": [] }, { "id": "yunet", "name": "YuNet", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "microcontroller" ], "footprint_note": "Edge or toolkit builds listed in: STM32 model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Millisecond-level face detector from OpenCV Zoo (Shiqi Yu et al.), ported to STM32 microcontrollers.", "verified_on": "2026-10-12", "sources": [ "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/face_detection/yunet" ], "unverified": [] }, { "id": "alibaba-damo-damofd", "name": "DamoFD", "level": "model", "family_id": null, "developer": [ "alibaba-damo" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba DAMO Academy lightweight face detector found by architecture search, published on ModelScope.", "verified_on": "2026-10-12", "sources": [ "https://www.modelscope.cn/models/iic/cv_ddsar_face-detection_iclr23-damofd" ], "unverified": [] }, { "id": "alibaba-damo-mogface", "name": "MogFace", "level": "model", "family_id": null, "developer": [ "alibaba-damo" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba DAMO Academy high-accuracy face detector (CVPR 2022), published on ModelScope.", "verified_on": "2026-10-12", "sources": [ "https://www.modelscope.cn/models/iic/cv_resnet101_face-detection_cvpr22papermogface" ], "unverified": [] }, { "id": "inspireface", "name": "InspireFace", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "SDK model pack for face detection, recognition and attributes on edge devices.", "verified_on": "2026-10-12", "sources": [ "https://www.modelscope.cn/models/tunmxy/InspireFace" ], "unverified": [] }, { "id": "cavaface", "name": "CavaFace", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-04-15", "release_date_basis": "registry date", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open-source PyTorch face recognition training framework with pretrained models.", "verified_on": "2026-10-12", "sources": [ "https://github.com/cavalleria/cavaface", "https://aihub.qualcomm.com/models/cavaface" ], "unverified": [] }, { "id": "emotion-ferplus", "name": "Emotion FERPlus", "level": "model", "family_id": null, "developer": [ "microsoft-research" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2016-08-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Facial expression recognition CNN trained on crowd-sourced FER+ labels; in the ONNX Model Zoo.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1608.01041", "https://github.com/onnx/models/tree/main/validated/vision/body_analysis/emotion_ferplus", "https://arxiv.org/html/1608.01041" ], "unverified": [] }, { "id": "l2cs-net", "name": "L2CS-Net", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Gaze estimation network supported by Roboflow, used for driver and attention monitoring.", "verified_on": "2026-10-12", "sources": [ "https://docs.roboflow.com/models/supported-models/l2cs-net.md" ], "unverified": [ "summary" ] }, { "id": "osnet", "name": "OSNet", "level": "model", "family_id": null, "developer": [ "university-of-surrey", "queen-mary-university-of-london" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-05-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU", "microcontroller" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, STM32 model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Omni-scale network for person re-identification; STMicroelectronics ports it to STM32.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1905.00953", "https://github.com/KaiyangZhou/deep-person-reid", "https://aihub.qualcomm.com/models/osnet", "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/re_identification/osnet", "https://arxiv.org/html/1905.00953" ], "unverified": [] }, { "id": "tsn", "name": "TSN (Temporal Segment Networks)", "level": "model", "family_id": null, "developer": [ "eth-zurich", "cuhk" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2016-08-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Action recognition by sparse temporal sampling of video segments.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1608.00859", "https://github.com/open-mmlab/mmaction2/tree/main/configs/recognition/tsn", "https://arxiv.org/html/1608.00859" ], "unverified": [] }, { "id": "tsm", "name": "TSM (Temporal Shift Module)", "level": "model", "family_id": null, "developer": [ "massachusetts-institute-of-technology" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-11-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Efficient video recognition by shifting channels along time at zero extra cost.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1811.08383", "https://github.com/open-mmlab/mmaction2/tree/main/configs/recognition/tsm", "https://arxiv.org/html/1811.08383" ], "unverified": [] }, { "id": "meta-slowfast", "name": "SlowFast", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "SlowFast and SlowOnly (3D ResNet Slow) in PyTorchVideo.", "context_length": null, "release_date": "2018-12-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta video recognition network with a slow pathway for semantics and a fast pathway for motion.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1812.03982", "https://github.com/facebookresearch/pytorchvideo", "https://pytorch.org/hub/facebookresearch_pytorchvideo_slowfast", "https://raw.githubusercontent.com/pytorch/hub/master/facebookresearch_pytorchvideo_slowfast.md", "https://pytorch.org/hub/facebookresearch_pytorchvideo_resnet", "https://raw.githubusercontent.com/pytorch/hub/master/facebookresearch_pytorchvideo_resnet.md", "https://github.com/open-mmlab/mmaction2/tree/main/configs/recognition/slowfast" ], "unverified": [] }, { "id": "meta-x3d", "name": "X3D", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-04-09", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta efficient video recognition networks expanded along several axes from a tiny 2D base.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2004.04730", "https://github.com/facebookresearch/pytorchvideo", "https://pytorch.org/hub/facebookresearch_pytorchvideo_x3d", "https://raw.githubusercontent.com/pytorch/hub/master/facebookresearch_pytorchvideo_x3d.md", "https://github.com/open-mmlab/mmaction2/tree/main/configs/recognition/x3d" ], "unverified": [] }, { "id": "google-deepmind-i3d", "name": "I3D (Inflated 3D ConvNet)", "level": "model", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-05-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DeepMind two-stream inflated 3D ConvNet for action recognition trained on Kinetics.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1705.07750", "https://www.kaggle.com/models/deepmind/i3d-kinetics", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/i3d-rgb-tf", "https://github.com/tensorflow/tfhub.dev/tree/master/assets/docs/deepmind", "https://github.com/open-mmlab/mmaction2/tree/main/configs/recognition/i3d" ], "unverified": [] }, { "id": "c3d", "name": "C3D", "level": "model", "family_id": null, "developer": [ "meta", "dartmouth-college" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2014-12-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Spatiotemporal features learned with 3D convolutional networks.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1412.0767", "https://github.com/open-mmlab/mmaction2/tree/main/configs/recognition/c3d", "https://arxiv.org/html/1412.0767" ], "unverified": [ "modalities_out" ] }, { "id": "video-resnet", "name": "Video ResNet (R3D, MC3, R(2+1)D)", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-11-30", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Spatiotemporal ResNet variants for action recognition, from the paper \"A Closer Look at Spatiotemporal Convolutions for Action Recognition\"; in TorchVision.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1711.11248", "https://aihub.qualcomm.com/models/resnet_3d", "https://github.com/pytorch/vision/blob/main/torchvision/models/video/resnet.py", "https://github.com/open-mmlab/mmaction2/tree/main/configs/recognition/r2plus1d", "https://pytorch.org/vision/stable/models/video_resnet.html" ], "unverified": [] }, { "id": "csn", "name": "CSN", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-04-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Video classification with channel-separated convolutional networks.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1904.02811", "https://github.com/open-mmlab/mmaction2/tree/main/configs/recognition/csn", "https://arxiv.org/html/1904.02811" ], "unverified": [] }, { "id": "nonlocal-video", "name": "Non-local Networks (video)", "level": "model", "family_id": null, "developer": [ "carnegie-mellon-university", "meta" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-11-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Video recognition with non-local self-attention blocks.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1711.07971", "https://github.com/open-mmlab/mmaction2/tree/main/configs/recognition/c2d", "https://arxiv.org/html/1711.07971" ], "unverified": [ "developer" ] }, { "id": "s3d", "name": "S3D", "level": "model", "family_id": null, "developer": [ "google", "uc-san-diego" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-12-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Separable 3D convolution video classifier; the OpenVINO common sign language model is built on it.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1712.04851", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/common-sign-language-0001", "https://pytorch.org/vision/stable/models/video_s3d.html", "https://github.com/tensorflow/models/tree/master/official/projects/s3d", "https://arxiv.org/html/1712.04851" ], "unverified": [] }, { "id": "trn", "name": "TRN", "level": "model", "family_id": null, "developer": [ "massachusetts-institute-of-technology" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-11-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Temporal relational reasoning network for video.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1711.08496", "https://github.com/open-mmlab/mmaction2/tree/main/configs/recognition/trn", "https://arxiv.org/html/1711.08496" ], "unverified": [] }, { "id": "tin", "name": "TIN", "level": "model", "family_id": null, "developer": [ "tsinghua-university" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-01-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "microcontroller" ], "footprint_note": "Edge or toolkit builds listed in: STM32 model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Temporal interlacing network for video recognition.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2001.06499", "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/object_detection/yolov2t", "https://github.com/open-mmlab/mmaction2/tree/main/configs/recognition/tin", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://arxiv.org/abs/2207.10666", "https://www.kaggle.com/models/google/tiny-video-net", "https://github.com/onnx/models/tree/main/validated/vision/object_detection_segmentation/tiny-yolov2", "https://arxiv.org/html/2001.06499" ], "unverified": [ "sources" ] }, { "id": "tpn", "name": "TPN", "level": "model", "family_id": null, "developer": [ "cuhk", "sensetime" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-04-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Temporal pyramid network for action recognition.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2004.03548", "https://github.com/open-mmlab/mmaction2/tree/main/configs/recognition/tpn", "https://arxiv.org/html/2004.03548" ], "unverified": [] }, { "id": "tanet", "name": "TANet (TAM)", "level": "model", "family_id": null, "developer": [ "nanjing-university", "sensetime" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-05-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Video recognition with a temporal adaptive module.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2005.06803", "https://github.com/open-mmlab/mmaction2/tree/main/configs/recognition/tanet", "https://arxiv.org/html/2005.06803" ], "unverified": [] }, { "id": "google-movinet", "name": "MoViNet", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "A0 to A6, base and streaming.", "context_length": null, "release_date": "2021-03-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google mobile video networks for efficient, streaming video recognition.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.11511", "https://www.kaggle.com/models/google/movinet", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://github.com/tensorflow/models/blob/master/official/README.md" ], "unverified": [] }, { "id": "google-tiny-video-net", "name": "Tiny Video Net", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google small video classifiers found by architecture search, published on Kaggle.", "verified_on": "2026-10-12", "sources": [ "https://www.kaggle.com/models/google/tiny-video-net" ], "unverified": [] }, { "id": "google-assemblenet", "name": "AssembleNet", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "AssembleNet and AssembleNet++.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google multi-stream video architectures found by search, in the TensorFlow Model Garden.", "verified_on": "2026-10-12", "sources": [ "https://github.com/tensorflow/models/tree/master/official/projects/assemblenet" ], "unverified": [] }, { "id": "google-youtube-8m-classifiers", "name": "YouTube-8M video classifiers", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google video-level classifiers (DBoF and others) for the YouTube-8M dataset, in the TensorFlow Model Garden.", "verified_on": "2026-10-12", "sources": [ "https://github.com/tensorflow/models/tree/master/official/projects/yt8m" ], "unverified": [] }, { "id": "meta-timesformer", "name": "TimeSformer", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-02-09", "release_date_basis": "announcement page", "weights_date": "2022-10-07", "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta video transformer with divided space-time attention.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2102.05095", "https://huggingface.co/facebook/timesformer-base-finetuned-k400", "https://huggingface.co/api/models?pipeline_tag=video-classification&sort=downloads&limit=50", "https://github.com/open-mmlab/mmaction2/tree/main/configs/recognition/timesformer" ], "unverified": [] }, { "id": "google-vivit", "name": "ViViT", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-03-29", "release_date_basis": "announcement page", "weights_date": "2022-11-23", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google video vision transformer for video classification.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.15691", "https://huggingface.co/google/vivit-b-16x2-kinetics400", "https://huggingface.co/api/models?pipeline_tag=video-classification&sort=downloads&limit=50" ], "unverified": [] }, { "id": "videomae", "name": "VideoMAE", "level": "model", "family_id": null, "developer": [ "nanjing-university", "tencent" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": 94210944, "params_active": null, "params_checkpoint": null, "params_note": "VideoMAE small, base, large; VideoMAE V2 (arXiv 2303.16727) up to ViT-g.", "context_length": null, "release_date": "2022-03-23", "release_date_basis": "announcement page", "weights_date": "2022-08-03", "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Self-supervised video pretraining with masked autoencoders, from Nanjing University.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2203.12602", "https://huggingface.co/MCG-NJU/videomae-base", "https://aihub.qualcomm.com/models/video_mae", "https://github.com/MCG-NJU/VideoMAE", "https://huggingface.co/api/models?pipeline_tag=video-classification&sort=downloads&limit=50", "https://github.com/open-mmlab/mmaction2/tree/main/configs/recognition/videomae", "https://github.com/open-mmlab/mmaction2/tree/main/configs/recognition/videomaev2", "https://arxiv.org/html/2203.12602" ], "unverified": [ "modalities_out" ] }, { "id": "videomae-v2", "name": "VideoMAE V2", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 86227200, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-03-29", "release_date_basis": "announcement page", "weights_date": "2025-01-14", "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "VideoMAE scaled with dual masking to billion-parameter video models.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2303.16727", "https://huggingface.co/OpenGVLab/VideoMAEv2-Base", "https://github.com/open-mmlab/mmaction2/tree/main/configs/recognition/videomaev2" ], "unverified": [] }, { "id": "video-swin-transformer", "name": "Video Swin Transformer", "level": "model", "family_id": null, "developer": [ "microsoft-research" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-06-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Swin Transformer extended to video recognition.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2106.13230", "https://github.com/open-mmlab/mmaction2/tree/main/configs/recognition/swin", "https://pytorch.org/vision/stable/models/video_swin_transformer.html", "https://arxiv.org/html/2106.13230" ], "unverified": [] }, { "id": "uniformer", "name": "UniFormer", "level": "model", "family_id": null, "developer": [ "siat", "shanghai-ai-lab" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "UniFormer and UniFormerV2 (arXiv 2211.09552).", "context_length": null, "release_date": "2022-01-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Unified convolution and self-attention transformer for efficient image and video recognition.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2201.04676", "https://github.com/open-mmlab/mmaction2/tree/main/configs/recognition/uniformer", "https://github.com/open-mmlab/mmaction2/tree/main/configs/recognition/uniformerv2", "https://arxiv.org/abs/2211.09552", "https://arxiv.org/html/2201.04676" ], "unverified": [] }, { "id": "microsoft-x-clip", "name": "X-CLIP", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 196585855, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-08-04", "release_date_basis": "announcement page", "weights_date": "2022-08-25", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft video-text model that extends CLIP to video recognition and zero-shot video classification.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2208.02816", "https://huggingface.co/microsoft/xclip-base-patch32", "https://huggingface.co/api/models?pipeline_tag=video-classification&sort=downloads&limit=50" ], "unverified": [ "developer" ] }, { "id": "actionclip", "name": "ActionCLIP", "level": "model", "family_id": null, "developer": [ "zhejiang-university" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-09-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Video action recognition framed as video-text matching.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2109.08472", "https://github.com/open-mmlab/mmaction2/tree/main/projects/actionclip", "https://arxiv.org/html/2109.08472" ], "unverified": [] }, { "id": "google-videoprism", "name": "VideoPrism", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-02-20", "release_date_basis": "announcement page", "weights_date": "2025-06-14", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google foundational video encoder for classification, retrieval and captioning.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2402.13217", "https://huggingface.co/google/videoprism-base-f16r288", "https://huggingface.co/api/models?pipeline_tag=video-classification&sort=likes&limit=50", "https://www.kaggle.com/models/keras/videoprism", "https://keras.io/keras_hub/presets/" ], "unverified": [ "modalities_out" ] }, { "id": "st-gcn", "name": "ST-GCN", "level": "model", "family_id": null, "developer": [ "cuhk" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "GNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-01-23", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Spatial temporal graph convolutional network for skeleton-based action recognition.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1801.07455", "https://github.com/open-mmlab/mmaction2/tree/main/configs/skeleton/stgcn", "https://github.com/open-mmlab/mmaction2/tree/main/configs/skeleton/stgcnpp", "https://arxiv.org/abs/2205.09443", "https://arxiv.org/html/1801.07455" ], "unverified": [] }, { "id": "2s-agcn", "name": "2s-AGCN", "level": "model", "family_id": null, "developer": [ "casia" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "GNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-05-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Two-stream adaptive graph convolutional network for skeleton-based action recognition.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1805.07694", "https://github.com/open-mmlab/mmaction2/tree/main/configs/skeleton/2s-agcn", "https://arxiv.org/html/1805.07694" ], "unverified": [] }, { "id": "ms-g3d", "name": "MS-G3D", "level": "model", "family_id": null, "developer": [ "university-of-sydney", "sensetime" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "GNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-03-31", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Skeleton action recognition with disentangled multi-scale graph convolutions.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2003.14111", "https://github.com/open-mmlab/mmaction2/tree/main/projects/msg3d", "https://arxiv.org/html/2003.14111" ], "unverified": [] }, { "id": "ctr-gcn", "name": "CTR-GCN", "level": "model", "family_id": null, "developer": [ "casia" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "GNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-07-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Channel-wise topology refinement graph convolution for skeleton action recognition.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2107.12213", "https://github.com/open-mmlab/mmaction2/tree/main/projects/ctrgcn", "https://arxiv.org/html/2107.12213" ], "unverified": [] }, { "id": "posec3d", "name": "PoseC3D", "level": "model", "family_id": null, "developer": [ "cuhk", "shanghai-ai-lab" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-04-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Skeleton action recognition with 3D CNNs on pose heatmaps, from the paper \"Revisiting Skeleton-based Action Recognition\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2104.13586", "https://github.com/open-mmlab/mmaction2/tree/main/configs/skeleton/posec3d", "https://arxiv.org/html/2104.13586" ], "unverified": [] }, { "id": "bsn", "name": "BSN", "level": "model", "family_id": null, "developer": [ "shanghai-jiao-tong-university" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-06-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Boundary sensitive network for temporal action proposal generation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1806.02964", "https://github.com/open-mmlab/mmaction2/tree/main/configs/localization/bsn", "https://arxiv.org/html/1806.02964" ], "unverified": [] }, { "id": "bmn", "name": "BMN", "level": "model", "family_id": null, "developer": [ "baidu" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-07-23", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Boundary-matching network for temporal action proposal generation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1907.09702", "https://github.com/open-mmlab/mmaction2/tree/main/configs/localization/bmn", "https://arxiv.org/html/1907.09702" ], "unverified": [] }, { "id": "ssn", "name": "SSN (Structured Segment Networks)", "level": "model", "family_id": null, "developer": [ "cuhk", "eth-zurich" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-04-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Temporal action detection with structured segment networks.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1704.06228", "https://github.com/open-mmlab/mmaction/tree/master/", "https://arxiv.org/html/1704.06228" ], "unverified": [] }, { "id": "tcanet", "name": "TCANet", "level": "model", "family_id": null, "developer": [ "hust", "sensetime" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-03-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Temporal context aggregation network for action proposal refinement.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.13141", "https://github.com/open-mmlab/mmaction2/tree/main/configs/localization/tcanet", "https://arxiv.org/html/2103.13141" ], "unverified": [] }, { "id": "acrn", "name": "ACRN", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-07-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Actor-centric relation network for spatio-temporal action detection.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1807.10982", "https://github.com/open-mmlab/mmaction2/tree/main/configs/detection/acrn", "https://arxiv.org/html/1807.10982" ], "unverified": [] }, { "id": "lfb", "name": "LFB (Long-Term Feature Bank)", "level": "model", "family_id": null, "developer": [ "ut-austin", "meta" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-12-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Video understanding with a long-term feature bank for detailed action detection.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1812.05038", "https://github.com/open-mmlab/mmaction2/tree/main/configs/detection/lfb", "https://arxiv.org/html/1812.05038" ], "unverified": [ "modalities_out" ] }, { "id": "dff", "name": "DFF (Deep Feature Flow)", "level": "model", "family_id": null, "developer": [ "university-of-science-and-technology-of-china", "microsoft-research" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2016-11-23", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Video recognition that propagates deep features between key frames with optical flow.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1611.07715", "https://github.com/open-mmlab/mmtracking/tree/master/configs/vid/dff", "https://arxiv.org/html/1611.07715" ], "unverified": [ "modalities_out" ] }, { "id": "fgfa", "name": "FGFA", "level": "model", "family_id": null, "developer": [ "university-of-science-and-technology-of-china", "microsoft-research" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-03-29", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Flow-guided feature aggregation for video object detection.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1703.10025", "https://github.com/open-mmlab/mmtracking/tree/master/configs/vid/fgfa", "https://arxiv.org/html/1703.10025" ], "unverified": [ "modalities_out" ] }, { "id": "selsa", "name": "SELSA", "level": "model", "family_id": null, "developer": [ "mcgill-university", "tusimple" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-07-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Sequence level semantics aggregation for video object detection.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1907.06390", "https://github.com/open-mmlab/mmtracking/tree/master/configs/vid/selsa", "https://arxiv.org/html/1907.06390" ], "unverified": [] }, { "id": "masktrack-r-cnn", "name": "MaskTrack R-CNN", "level": "model", "family_id": null, "developer": [ "bytedance", "uiuc", "adobe" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-05-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Video instance segmentation baseline, from the paper \"Video Instance Segmentation\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1905.04804", "https://github.com/open-mmlab/mmtracking/tree/master/configs/vis/masktrack_rcnn", "https://arxiv.org/html/1905.04804" ], "unverified": [] }, { "id": "sort", "name": "SORT", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2016-02-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Simple online and realtime multi-object tracking with a Kalman filter and the Hungarian algorithm.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1602.00763", "https://github.com/open-mmlab/mmtracking/tree/master/configs/mot/deepsort" ], "unverified": [] }, { "id": "deepsort", "name": "DeepSORT", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-03-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "SORT extended with a deep appearance descriptor for re-identification across occlusions.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1703.07402", "https://modelzoo.co/model/yolov4-deepsort", "https://github.com/theAIGuysCode/yolov4-deepsort", "https://github.com/open-mmlab/mmtracking/tree/master/configs/mot/deepsort" ], "unverified": [] }, { "id": "strongsort", "name": "StrongSORT", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-02-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Upgraded DeepSORT, from the paper \"StrongSORT: Make DeepSORT Great Again\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2202.13514", "https://github.com/open-mmlab/mmtracking/tree/master/configs/mot/strongsort" ], "unverified": [] }, { "id": "bytetrack", "name": "ByteTrack", "level": "model", "family_id": null, "developer": [ "hust" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-10-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multi-object tracker that associates every detection box, including low-score ones; deployable from Vertex AI and Azure AI Foundry.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2110.06864", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden/model_garden_video_object_tracking_serve.ipynb", "https://ai.azure.com/catalog/models?task=multi-object-tracking", "https://www.modelscope.cn/models/AIBS/bytetrack", "https://github.com/open-mmlab/mmtracking/tree/master/configs/mot/bytetrack", "https://arxiv.org/html/2110.06864" ], "unverified": [] }, { "id": "oc-sort", "name": "OC-SORT", "level": "model", "family_id": null, "developer": [ "carnegie-mellon-university", "shanghai-ai-lab", "nvidia" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-03-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Observation-centric SORT for robust multi-object tracking.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2203.14360", "https://ai.azure.com/catalog/models?task=object-detection", "https://github.com/open-mmlab/mmtracking/tree/master/configs/mot/ocsort", "https://arxiv.org/html/2203.14360" ], "unverified": [ "developer" ] }, { "id": "fairmot", "name": "FairMOT", "level": "model", "family_id": null, "developer": [ "hust", "microsoft-research" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-04-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "One-shot multi-object tracker that balances detection and re-identification.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2004.01888", "https://www.modelscope.cn/models/iic/cv_yolov5_video-multi-object-tracking_fairmot", "https://arxiv.org/html/2004.01888" ], "unverified": [] }, { "id": "qdtrack", "name": "QDTrack", "level": "model", "family_id": null, "developer": [ "zhejiang-university", "eth-zurich" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-06-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multi-object tracking with quasi-dense similarity learning.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2006.06664", "https://github.com/open-mmlab/mmtracking/tree/master/configs/mot/qdtrack", "https://arxiv.org/html/2006.06664" ], "unverified": [] }, { "id": "tracktor", "name": "Tracktor", "level": "model", "family_id": null, "developer": [ "tum" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-03-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tracker that uses a detector's box regressor to follow objects, from the paper \"Tracking without bells and whistles\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1903.05625", "https://github.com/open-mmlab/mmtracking/tree/master/configs/mot/tracktor", "https://arxiv.org/html/1903.05625" ], "unverified": [] }, { "id": "siamrpn-plus-plus", "name": "SiamRPN++", "level": "model", "family_id": null, "developer": [ "sensetime", "casia" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-12-31", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Siamese single-object tracker with very deep networks.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1812.11703", "https://github.com/open-mmlab/mmtracking/tree/master/configs/sot/siamese_rpn", "https://arxiv.org/html/1812.11703" ], "unverified": [] }, { "id": "stark", "name": "STARK", "level": "model", "family_id": null, "developer": [ "dalian-university-of-technology", "microsoft-research" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-03-31", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Spatio-temporal transformer for single-object visual tracking.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.17154", "https://github.com/open-mmlab/mmtracking/tree/master/configs/sot/stark", "https://arxiv.org/html/2103.17154" ], "unverified": [] }, { "id": "mixformer", "name": "MixFormer", "level": "model", "family_id": null, "developer": [ "nanjing-university" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-03-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "End-to-end tracking with iterative mixed attention.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2203.11082", "https://github.com/open-mmlab/mmtracking/tree/master/configs/sot/mixformer", "https://arxiv.org/html/2203.11082" ], "unverified": [] }, { "id": "prdimp", "name": "PrDiMP", "level": "model", "family_id": null, "developer": [ "eth-zurich" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-03-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Probabilistic regression for single-object visual tracking.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2003.12565", "https://github.com/open-mmlab/mmtracking/tree/master/configs/sot/prdimp", "https://arxiv.org/html/2003.12565" ], "unverified": [] }, { "id": "ostrack", "name": "OSTrack", "level": "model", "family_id": null, "developer": [ "ict-cas", "university-of-chinese-academy-of-sciences" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-03-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "One-stream transformer tracker with joint feature learning and relation modelling.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2203.11991", "https://www.modelscope.cn/models/iic/cv_vitb_video-single-object-tracking_ostrack", "https://arxiv.org/html/2203.11991" ], "unverified": [ "modalities_out" ] }, { "id": "crnn", "name": "CRNN", "level": "model", "family_id": null, "developer": [ "hust" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2015-07-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Convolutional recurrent network for image-based sequence and scene text recognition; the base of many OCR recognisers (keras-ocr, PaddleOCR, ModelScope).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1507.05717", "https://github.com/bgshih/crnn", "https://www.modelscope.cn/models/iic/cv_crnn_ocr-recognition-general_damo", "https://modelzoo.co/model/an-end-to-end-trainable-neural-network-for", "https://www.kaggle.com/models/kaggle/keras-ocr", "https://www.kaggle.com/models/kaggle/rosetta", "https://github.com/tensorflow/tfhub.dev/tree/master/assets/docs/tulasiram58827", "https://arxiv.org/html/1507.05717" ], "unverified": [] }, { "id": "east-text-detector", "name": "EAST text detector", "level": "model", "family_id": null, "developer": [ "megvii" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-04-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Efficient and accurate scene text detector; widely used through OpenCV.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1704.03155", "https://github.com/argman/EAST", "https://modelzoo.co/model/east-tensorflow", "https://www.kaggle.com/models/spsayakpaul/east-text-detector", "https://arxiv.org/html/1704.03155" ], "unverified": [] }, { "id": "naver-craft", "name": "CRAFT text detector", "level": "model", "family_id": null, "developer": [ "naver" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-04-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NAVER Clova character-region text detector; the detector inside EasyOCR.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1904.01941", "https://github.com/clovaai/CRAFT-pytorch", "https://modelzoo.co/model/craft-pytorch", "https://www.kaggle.com/models/kaggle/craft-text-detector", "https://github.com/tensorflow/tfhub.dev/tree/master/assets/docs/tulasiram58827" ], "unverified": [] }, { "id": "ctpn", "name": "CTPN", "level": "model", "family_id": null, "developer": [ "siat" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Connectionist text proposal network for horizontal text detection, in the OpenVINO Open Model Zoo.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/ctpn", "https://arxiv.org/html/1609.03605" ], "unverified": [ "modalities_out" ] }, { "id": "dbnet", "name": "DBNet", "level": "model", "family_id": null, "developer": [ "hust" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "DBNet and DBNet++ (arXiv 2202.10304); ModelScope line-level DAMO build.", "context_length": null, "release_date": "2019-11-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time scene text detector with differentiable binarisation; the detector used in PaddleOCR and MMOCR.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1911.08947", "https://www.modelscope.cn/models/iic/cv_resnet18_ocr-detection-db-line-level_damo", "https://github.com/open-mmlab/mmocr/tree/main/configs/textdet/dbnet", "https://github.com/open-mmlab/mmocr/tree/main/configs/textdet/dbnetpp", "https://arxiv.org/abs/2202.10304", "https://arxiv.org/html/1911.08947" ], "unverified": [] }, { "id": "psenet", "name": "PSENet", "level": "model", "family_id": null, "developer": [ "nanjing-university" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-03-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Shape-robust text detector with progressive scale expansion.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1903.12473", "https://github.com/open-mmlab/mmocr/tree/main/configs/textdet/psenet", "https://arxiv.org/html/1903.12473" ], "unverified": [] }, { "id": "panet-text", "name": "PANet (text)", "level": "model", "family_id": null, "developer": [ "nanjing-university" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-08-16", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Arbitrary-shaped text detector with a pixel aggregation network.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1908.05900", "https://github.com/open-mmlab/mmocr/tree/main/configs/textdet/panet", "https://arxiv.org/html/1908.05900" ], "unverified": [] }, { "id": "textsnake", "name": "TextSnake", "level": "model", "family_id": null, "developer": [ "peking-university", "megvii" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-07-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Flexible representation for detecting text of arbitrary shapes.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1807.01544", "https://github.com/open-mmlab/mmocr/tree/main/configs/textdet/textsnake", "https://arxiv.org/html/1807.01544" ], "unverified": [] }, { "id": "drrg", "name": "DRRG", "level": "model", "family_id": null, "developer": [ "ustb" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "GNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-03-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Arbitrary-shape text detection with a relational reasoning graph network.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2003.07493", "https://github.com/open-mmlab/mmocr/tree/main/configs/textdet/drrg", "https://arxiv.org/html/2003.07493" ], "unverified": [] }, { "id": "fcenet", "name": "FCENet", "level": "model", "family_id": null, "developer": [ "scut", "sensetime" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-04-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Arbitrary-shaped text detection with Fourier contour embedding.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2104.10442", "https://github.com/open-mmlab/mmocr/tree/main/configs/textdet/fcenet", "https://arxiv.org/html/2104.10442" ], "unverified": [] }, { "id": "abcnet", "name": "ABCNet", "level": "model", "family_id": null, "developer": [ "scut", "university-of-adelaide" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "ABCNet and ABCNet v2 (arXiv 2105.03620).", "context_length": null, "release_date": "2020-02-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time end-to-end text spotting with adaptive Bezier-curve networks.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2002.10200", "https://github.com/open-mmlab/mmocr/tree/main/projects/ABCNet", "https://arxiv.org/abs/2105.03620", "https://arxiv.org/html/2002.10200" ], "unverified": [] }, { "id": "spts", "name": "SPTS", "level": "model", "family_id": null, "developer": [ "scut" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-12-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Text spotting with single-point annotations.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2112.07917", "https://github.com/open-mmlab/mmocr/tree/main/projects/SPTS", "https://arxiv.org/html/2112.07917" ], "unverified": [] }, { "id": "oclip", "name": "oCLIP", "level": "model", "family_id": null, "developer": [ "ntu-singapore", "bytedance" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-03-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Weakly supervised vision-language pretraining for scene text detection and spotting.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2203.03911", "https://github.com/open-mmlab/mmocr/tree/main/configs/backbone/oclip", "https://arxiv.org/html/2203.03911" ], "unverified": [] }, { "id": "abinet", "name": "ABINet", "level": "model", "family_id": null, "developer": [ "university-of-science-and-technology-of-china" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-03-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Scene text recogniser with autonomous, bidirectional and iterative language modelling.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.06495", "https://github.com/open-mmlab/mmocr/tree/main/configs/textrecog/abinet", "https://arxiv.org/html/2103.06495" ], "unverified": [] }, { "id": "aster", "name": "ASTER", "level": "model", "family_id": null, "developer": [], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Attentional scene text recogniser with flexible rectification (TPAMI 2018), reproduced in MMOCR.", "verified_on": "2026-10-12", "sources": [ "https://github.com/open-mmlab/mmocr/tree/main/configs/textrecog/aster" ], "unverified": [] }, { "id": "master-ocr", "name": "MASTER", "level": "model", "family_id": null, "developer": [ "ping-an" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-10-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multi-aspect non-local network for scene text recognition.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1910.02562", "https://github.com/open-mmlab/mmocr/tree/main/configs/textrecog/master", "https://arxiv.org/html/1910.02562" ], "unverified": [] }, { "id": "nrtr", "name": "NRTR", "level": "model", "family_id": null, "developer": [ "casia" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-06-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "No-recurrence sequence-to-sequence model for scene text recognition.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1806.00926", "https://github.com/open-mmlab/mmocr/tree/main/configs/textrecog/nrtr", "https://arxiv.org/html/1806.00926" ], "unverified": [] }, { "id": "robustscanner", "name": "RobustScanner", "level": "model", "family_id": null, "developer": [ "sensetime" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-07-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Text recogniser that strengthens positional clues.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2007.07542", "https://github.com/open-mmlab/mmocr/tree/main/configs/textrecog/robust_scanner", "https://arxiv.org/html/2007.07542" ], "unverified": [] }, { "id": "sar-ocr", "name": "SAR (Show, Attend and Read)", "level": "model", "family_id": null, "developer": [ "university-of-adelaide" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-11-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Baseline for irregular scene text recognition.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1811.00751", "https://github.com/open-mmlab/mmocr/tree/main/configs/textrecog/sar", "https://arxiv.org/html/1811.00751" ], "unverified": [] }, { "id": "satrn", "name": "SATRN", "level": "model", "family_id": null, "developer": [ "naver" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-10-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Recogniser for texts of arbitrary shapes with 2D self-attention.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1910.04396", "https://github.com/open-mmlab/mmocr/tree/main/configs/textrecog/satrn", "https://arxiv.org/html/1910.04396" ], "unverified": [] }, { "id": "svtr", "name": "SVTR", "level": "model", "family_id": null, "developer": [ "beijing-jiaotong-university", "baidu" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-04-30", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Scene text recognition with a single visual model; used in PP-OCRv3 and later.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2205.00159", "https://github.com/open-mmlab/mmocr/tree/main/configs/textrecog/svtr", "https://arxiv.org/html/2205.00159" ], "unverified": [ "summary" ] }, { "id": "parseq", "name": "PARSeq", "level": "model", "family_id": null, "developer": [ "university-of-the-philippines" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-07-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Scene text recognition with permuted autoregressive sequence models; in KerasHub.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2207.06966", "https://www.kaggle.com/models/keras/parseq", "https://keras.io/keras_hub/presets/", "https://arxiv.org/html/2207.06966" ], "unverified": [] }, { "id": "alibaba-damo-mgp-str", "name": "MGP-STR", "level": "model", "family_id": null, "developer": [ "alibaba-damo" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 147952305, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-09-08", "release_date_basis": "announcement page", "weights_date": "2022-11-23", "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba DAMO Academy scene text recogniser with multi-granularity prediction.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2209.03592", "https://huggingface.co/alibaba-damo/mgp-str-base", "https://huggingface.co/api/models?pipeline_tag=image-to-text&sort=downloads&limit=50" ], "unverified": [] }, { "id": "attention-ocr", "name": "Attention OCR", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-04-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Attention-based extraction of structured text from Street View imagery.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1704.03549", "https://github.com/tensorflow/models/blob/master/research/README.md", "https://arxiv.org/html/1704.03549" ], "unverified": [] }, { "id": "sdmg-r", "name": "SDMG-R", "level": "model", "family_id": null, "developer": [ "sensetime" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "GNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-03-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Key information extraction from documents with spatial dual-modality graph reasoning; in MMOCR.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.14470", "https://github.com/open-mmlab/mmocr/tree/main/configs/kie/sdmgr", "https://arxiv.org/html/2103.14470" ], "unverified": [] }, { "id": "easyocr", "name": "EasyOCR", "level": "model", "family_id": null, "developer": [ "jaided-ai" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-03-14", "release_date_basis": "registry date", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Ready-to-use OCR library supporting 80 plus languages, built on CRAFT detection and CRNN recognition.", "verified_on": "2026-10-12", "sources": [ "https://github.com/JaidedAI/EasyOCR", "https://aihub.qualcomm.com/models/easyocr", "https://docs.roboflow.com/models/supported-models/easyocr.md" ], "unverified": [] }, { "id": "doctr", "name": "docTR", "level": "model", "family_id": null, "developer": [], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open-source document OCR library from Mindee with detection and recognition models; a Roboflow supported model.", "verified_on": "2026-10-12", "sources": [ "https://docs.roboflow.com/models/supported-models/doctr.md" ], "unverified": [] }, { "id": "baidu-pp-ocr", "name": "PP-OCR", "level": "family", "family_id": null, "developer": [ "baidu" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "PP-OCRv5 server and mobile detection and recognition, PP-OCRv6 medium (arXiv 2606.13108), and helpers: PP-LCNet document and text-line orientation, UVDoc unwarping.", "context_length": null, "release_date": "2025-06-04", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Baidu PaddlePaddle family of practical OCR pipelines with text detection, recognition and orientation models; widely used for multilingual document OCR.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/PaddlePaddle/PP-OCRv5_server_det", "https://huggingface.co/api/models?pipeline_tag=image-to-text&sort=downloads&limit=50" ], "unverified": [] }, { "id": "baidu-pp-ocrv5-server", "name": "PP-OCRv5 server", "level": "variant", "family_id": "baidu-pp-ocr", "developer": [ "baidu" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-06-04", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Server-grade PP-OCRv5 text detection and recognition models.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/PaddlePaddle/PP-OCRv5_server_rec" ], "unverified": [] }, { "id": "baidu-pp-ocrv5-mobile", "name": "PP-OCRv5 mobile", "level": "variant", "family_id": "baidu-pp-ocr", "developer": [ "baidu" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-06-04", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Lightweight PP-OCRv5 models for mobile and edge devices.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/PaddlePaddle/PP-OCRv5_mobile_det" ], "unverified": [] }, { "id": "baidu-pp-ocrv6", "name": "PP-OCRv6", "level": "variant", "family_id": "baidu-pp-ocr", "developer": [ "baidu" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "1.5M to 34.5M parameters.", "context_length": null, "release_date": "2026-06-11", "release_date_basis": "announcement page", "weights_date": "2026-06-10", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "PP-OCRv6, from the paper \"PP-OCRv6: From 1.5M to 34.5M Parameters, Surpassing Billion-Scale VLMs on OCR Tasks\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2606.13108", "https://huggingface.co/PaddlePaddle/PP-OCRv6_medium_rec" ], "unverified": [] }, { "id": "baidu-pp-doclayoutv3", "name": "PP-DocLayoutV3", "level": "model", "family_id": null, "developer": [ "baidu" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01-20", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Baidu PaddlePaddle document layout analysis model that detects layout regions and reading order.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/PaddlePaddle/PP-DocLayoutV3", "https://huggingface.co/api/models?pipeline_tag=image-segmentation&sort=downloads&limit=50" ], "unverified": [] }, { "id": "baidu-unlimited-ocr", "name": "Unlimited OCR", "level": "model", "family_id": null, "developer": [ "baidu" ], "model_class": "OCR and document", "modalities_in": [ "image", "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 3336106240, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-06-22", "release_date_basis": "announcement page", "weights_date": "2026-06-19", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Baidu end-to-end document OCR vision-language model, from the paper \"Unlimited OCR Works\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2606.23050", "https://huggingface.co/baidu/Unlimited-OCR", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://huggingface.co/api/models?pipeline_tag=image-text-to-text&sort=likes&limit=50" ], "unverified": [] }, { "id": "alibaba-damo-ocr-models", "name": "DAMO OCR models (ModelScope)", "level": "model", "family_id": null, "developer": [ "alibaba-damo" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "ConvNeXt-Tiny general text recogniser, DBNet line-level detector and CRNN recogniser.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba DAMO Academy general Chinese and English OCR detection and recognition models on ModelScope.", "verified_on": "2026-10-12", "sources": [ "https://www.modelscope.cn/models/iic/cv_convnextTiny_ocr-recognition-general_damo", "https://www.modelscope.cn/models/iic/cv_resnet18_ocr-detection-db-line-level_damo", "https://www.modelscope.cn/models/iic/cv_crnn_ocr-recognition-general_damo" ], "unverified": [] }, { "id": "alibaba-damo-lore", "name": "LORE table structure recognition", "level": "model", "family_id": null, "developer": [ "alibaba-damo" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Cycle-CenterNet wired-table model is a related DAMO table recogniser.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba DAMO Academy table structure recognition model that predicts logical locations of cells, published on ModelScope.", "verified_on": "2026-10-12", "sources": [ "https://www.modelscope.cn/models/iic/cv_resnet-transformer_table-structure-recognition_lore", "https://www.modelscope.cn/models/iic/cv_dla34_table-structure-recognition_cycle-centernet" ], "unverified": [] }, { "id": "microsoft-trocr", "name": "TrOCR", "level": "family", "family_id": null, "developer": [ "microsoft" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Small, base and large for printed and handwritten text, plus stage-1 pretrained weights.", "context_length": null, "release_date": "2021-09-21", "release_date_basis": "announcement page", "weights_date": "2022-03-02", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft transformer OCR that pairs a pretrained image encoder with a pretrained text decoder for text-line recognition.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2109.10282", "https://huggingface.co/microsoft/trocr-base-handwritten", "https://aihub.qualcomm.com/models/trocr", "https://huggingface.co/microsoft/trocr-small-handwritten", "https://huggingface.co/api/models?pipeline_tag=image-to-text&sort=downloads&limit=50", "https://docs.roboflow.com/models/supported-models/trocr.md", "https://paperswithcode.com/method/trocr" ], "unverified": [] }, { "id": "microsoft-trocr-small-handwritten", "name": "TrOCR small handwritten", "level": "variant", "family_id": "microsoft-trocr", "developer": [ "microsoft" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-09-21", "release_date_basis": "announcement page", "weights_date": "2022-03-02", "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Smallest TrOCR fine-tuned on IAM handwriting.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2109.10282", "https://huggingface.co/microsoft/trocr-small-handwritten" ], "unverified": [] }, { "id": "microsoft-trocr-large-printed", "name": "TrOCR large printed", "level": "variant", "family_id": "microsoft-trocr", "developer": [ "microsoft" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 608119809, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-09-21", "release_date_basis": "announcement page", "weights_date": "2022-03-02", "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Largest TrOCR fine-tuned on printed receipts (SROIE).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2109.10282", "https://huggingface.co/microsoft/trocr-large-printed" ], "unverified": [] }, { "id": "microsoft-layoutlm", "name": "LayoutLM", "level": "family", "family_id": null, "developer": [ "microsoft" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "LayoutLM v1, v2, v3 and the multilingual LayoutXLM.", "context_length": null, "release_date": "2019-12-31", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft document AI models pretrained jointly on text, layout and image for forms, receipts and invoice understanding.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1912.13318", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://huggingface.co/impira/layoutlm-document-qa", "https://huggingface.co/api/models?pipeline_tag=document-question-answering&sort=downloads&limit=50", "https://huggingface.co/microsoft/layoutlmv3-base" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "microsoft-layoutlmv2", "name": "LayoutLMv2", "level": "variant", "family_id": "microsoft-layoutlm", "developer": [ "microsoft" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-12-29", "release_date_basis": "announcement page", "weights_date": "2022-03-02", "openness": "open weights restricted", "licence": "cc-by-nc-sa-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Second LayoutLM with visual features in pretraining.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2012.14740", "https://huggingface.co/microsoft/layoutlmv2-base-uncased" ], "unverified": [] }, { "id": "microsoft-layoutlmv3", "name": "LayoutLMv3", "level": "variant", "family_id": "microsoft-layoutlm", "developer": [ "microsoft" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 125327490, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-04-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-sa-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Third LayoutLM with unified text and image masking; the base for many invoice and form extractors.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2204.08387", "https://huggingface.co/microsoft/layoutlmv3-base" ], "unverified": [] }, { "id": "microsoft-layoutxlm", "name": "LayoutXLM", "level": "variant", "family_id": "microsoft-layoutlm", "developer": [ "microsoft" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-04-18", "release_date_basis": "announcement page", "weights_date": "2022-03-02", "openness": "open weights restricted", "licence": "cc-by-nc-sa-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multilingual LayoutLMv2 for visually rich documents in many languages.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2104.08836", "https://huggingface.co/microsoft/layoutxlm-base" ], "unverified": [] }, { "id": "impira-layoutlm-document-qa", "name": "LayoutLM document QA (Impira)", "level": "variant", "family_id": "microsoft-layoutlm", "developer": [], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 127793412, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-08-07", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "LayoutLM fine-tuned by Impira for question answering over document images; among the most downloaded document QA models on Hugging Face.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/impira/layoutlm-document-qa", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden" ], "unverified": [] }, { "id": "microsoft-dit", "name": "DiT (Document Image Transformer)", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-03-04", "release_date_basis": "announcement page", "weights_date": "2022-03-07", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft self-supervised document image transformer for document classification and layout analysis.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2203.02378", "https://huggingface.co/microsoft/dit-base-finetuned-rvlcdip", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://ai.azure.com/catalog/models?task=image-classification" ], "unverified": [] }, { "id": "microsoft-table-transformer", "name": "Table Transformer", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 28818631, "params_active": null, "params_checkpoint": null, "params_note": "Table detection and table structure recognition (v1.0 and v1.1).", "context_length": null, "release_date": "2021-09-30", "release_date_basis": "announcement page", "weights_date": "2022-10-14", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft DETR models for detecting tables and recognising their structure, trained on PubTables-1M.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2110.00061", "https://huggingface.co/microsoft/table-transformer-detection", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://huggingface.co/api/models?pipeline_tag=object-detection&sort=downloads&limit=50", "https://huggingface.co/microsoft/table-transformer-structure-recognition", "https://www.modelscope.cn/models/AI-ModelScope/table-transformer-structure-recognition-v1.1-all" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "naver-donut", "name": "Donut", "level": "model", "family_id": null, "developer": [ "naver" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-11-30", "release_date_basis": "announcement page", "weights_date": "2022-07-19", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NAVER Clova OCR-free document understanding transformer used for receipts, forms and document question answering.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2111.15664", "https://huggingface.co/naver-clova-ix/donut-base", "https://huggingface.co/api/models?pipeline_tag=image-to-text&sort=likes&limit=50", "https://www.modelscope.cn/models/cubeai/donut-base-finetuned-cord-v1" ], "unverified": [] }, { "id": "meta-nougat", "name": "Nougat", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "OCR and document", "modalities_in": [ "image", "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 348736012, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-08-25", "release_date_basis": "announcement page", "weights_date": "2023-09-21", "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta model that converts academic PDF pages to markup, including equations and tables.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2308.13418", "https://huggingface.co/facebook/nougat-base", "https://huggingface.co/api/models?pipeline_tag=image-to-text&sort=likes&limit=50", "https://www.modelscope.cn/models/cubeai/nougat-base" ], "unverified": [] }, { "id": "google-pix2struct", "name": "Pix2Struct", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 282285696, "params_active": null, "params_checkpoint": null, "params_note": "Pix2Struct, DePlot (arXiv 2212.10505) and MatCha (arXiv 2212.09662).", "context_length": null, "release_date": "2022-10-07", "release_date_basis": "announcement page", "weights_date": "2023-03-21", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google image-to-text model pretrained by parsing screenshots, fine-tuned for document, chart and UI understanding.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2210.03347", "https://huggingface.co/google/pix2struct-docvqa-base", "https://huggingface.co/api/models?pipeline_tag=visual-question-answering&sort=downloads&limit=50", "https://huggingface.co/google/deplot" ], "unverified": [] }, { "id": "google-deplot", "name": "DePlot", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 282285696, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-12-20", "release_date_basis": "announcement page", "weights_date": "2023-04-03", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google plot-to-table model that converts chart images into data tables for reasoning.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2212.10505", "https://huggingface.co/google/deplot", "https://www.kaggle.com/models/google/deplot" ], "unverified": [] }, { "id": "google-unified-detector", "name": "Unified Detector (text and layout)", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google model for joint text detection and layout analysis, in the TensorFlow Model Garden.", "verified_on": "2026-10-12", "sources": [ "https://github.com/tensorflow/models/tree/master/official/projects/unified_detector" ], "unverified": [] }, { "id": "opendatalab-pdf-extract-kit", "name": "PDF-Extract-Kit 1.0", "level": "model", "family_id": null, "developer": [ "opendatalab" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenDataLab bundle of layout detection, formula detection and recognition, table and OCR models used by MinerU for PDF extraction.", "verified_on": "2026-10-12", "sources": [ "https://www.modelscope.cn/models/OpenDataLab/PDF-Extract-Kit-1.0" ], "unverified": [] }, { "id": "alibaba-damo-mplug-docowl-1-5", "name": "mPLUG-DocOwl 1.5", "level": "model", "family_id": null, "developer": [ "alibaba-damo" ], "model_class": "OCR and document", "modalities_in": [ "image", "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba DAMO Academy OCR-free document understanding multimodal model, published on ModelScope.", "verified_on": "2026-10-12", "sources": [ "https://www.modelscope.cn/models/iic/DocOwl1.5-Chat" ], "unverified": [] }, { "id": "tencent-hunyuanocr", "name": "HunyuanOCR", "level": "model", "family_id": null, "developer": [ "tencent" ], "model_class": "OCR and document", "modalities_in": [ "image", "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tencent Hunyuan end-to-end OCR vision-language model, published on ModelScope.", "verified_on": "2026-10-12", "sources": [ "https://www.modelscope.cn/models/Tencent-Hunyuan/HunyuanOCR" ], "unverified": [] }, { "id": "baichuan-baichuanmed-ocr-7b", "name": "BaichuanMed OCR 7B", "level": "model", "family_id": null, "developer": [ "baichuan" ], "model_class": "OCR and document", "modalities_in": [ "image", "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8292166656, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-04-07", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Baichuan medical document OCR vision-language model with 7B parameters, published on Hugging Face.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/baichuan-inc/BaichuanMed-OCR-7B", "https://ai.azure.com/catalog/models?industry=health-and-life-sciences" ], "unverified": [] }, { "id": "lighton-lightonocr", "name": "LightOnOCR", "level": "model", "family_id": null, "developer": [ "lighton" ], "model_class": "OCR and document", "modalities_in": [ "image", "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 1005647872, "params_active": null, "params_checkpoint": null, "params_note": "LightOnOCR-1B (October 2025) and LightOnOCR-2-1B (January 2026).", "context_length": null, "release_date": "2025-10", "release_date_basis": "announcement page", "weights_date": "2026-01-16", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "LightOn 1B end-to-end multilingual vision-language model for document OCR.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2601.14251", "https://huggingface.co/lightonai/LightOnOCR-2-1B", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://huggingface.co/api/models?pipeline_tag=image-to-text&sort=likes&limit=50", "https://huggingface.co/lightonai/LightOnOCR-1B-1025" ], "unverified": [] }, { "id": "tii-falcon-ocr", "name": "Falcon-OCR", "level": "model", "family_id": null, "developer": [ "tii" ], "model_class": "OCR and document", "modalities_in": [ "image", "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 269944416, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-02-22", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "TII document OCR model released with Falcon Perception.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/tiiuae/Falcon-OCR", "https://huggingface.co/api/models?pipeline_tag=image-to-text&sort=likes&limit=50" ], "unverified": [] }, { "id": "nvidia-nemotron-ocr-v2", "name": "Nemotron OCR v2", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04-01", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA OCR model for text detection and recognition in document pipelines.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/nvidia/nemotron-ocr-v2", "https://huggingface.co/api/models?pipeline_tag=image-to-text&sort=likes&limit=50" ], "unverified": [] }, { "id": "nvidia-nemotron-parse", "name": "Nemotron Parse 1.1", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "OCR and document", "modalities_in": [ "image", "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 956758025, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-11-17", "release_date_basis": "announcement page", "weights_date": "2025-11-15", "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA document parsing model that extracts text, tables and layout with reading order.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2511.20478", "https://huggingface.co/nvidia/NVIDIA-Nemotron-Parse-v1.1", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "ibm-granite-vision-3-3-2b", "name": "Granite Vision 3.3 2B", "level": "model", "family_id": null, "developer": [ "ibm" ], "model_class": "OCR and document", "modalities_in": [ "image", "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 2975396928, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-06-11", "release_date_basis": "announcement page", "weights_date": "2025-06-03", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "IBM lightweight open vision-language model for enterprise document understanding such as tables, charts and forms.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2502.09927", "https://huggingface.co/ibm-granite/granite-vision-3.3-2b", "https://huggingface.co/api/models?pipeline_tag=image-to-text&sort=downloads&limit=50", "https://replicate.com/ibm-granite/granite-vision-3.3-2b" ], "unverified": [] }, { "id": "vintern-1b-v2", "name": "Vintern-1B v2", "level": "model", "family_id": null, "developer": [], "model_class": "OCR and document", "modalities_in": [ "image", "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 938158976, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-08-22", "release_date_basis": "announcement page", "weights_date": "2024-08-09", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vietnamese 1B multimodal model for OCR, document extraction and question answering on Vietnamese documents.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2408.12480", "https://huggingface.co/5CD-AI/Vintern-1B-v2", "https://huggingface.co/api/models?pipeline_tag=document-question-answering&sort=downloads&limit=50" ], "unverified": [] }, { "id": "datalab-surya-marker", "name": "Surya and Marker", "level": "model", "family_id": null, "developer": [ "datalab" ], "model_class": "OCR and document", "modalities_in": [ "image", "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Datalab OCR, layout and reading-order models (Surya) and the Marker converter built on them, which turns PDFs and other files into Markdown, JSON or HTML; code under Apache 2.0, weights under a modified AI Pubs Open Rail-M licence.", "verified_on": "2026-10-15", "sources": [ "https://github.com/datalab-to/marker", "https://github.com/datalab-to/surya", "https://replicate.com/datalab-to/marker" ] }, { "id": "bytedance-dolphin", "name": "Dolphin (document parsing)", "level": "model", "family_id": null, "developer": [ "bytedance" ], "model_class": "OCR and document", "modalities_in": [ "image", "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": null, "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ByteDance document image parsing model that first analyses page layout and reading order, then parses the elements; MIT licence, weights on Hugging Face.", "verified_on": "2026-10-15", "sources": [ "https://github.com/bytedance/Dolphin", "https://huggingface.co/ByteDance/Dolphin", "https://replicate.com/bytedance/dolphin" ], "unverified": [] }, { "id": "intel-text-and-formula-recognition", "name": "Intel OCR models (OpenVINO)", "level": "model", "family_id": null, "developer": [ "intel" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "horizontal-text-detection, text-recognition, handwritten English, Japanese and simplified Chinese recognition, formula recognition (medium-scan, polynomials handwritten).", "context_length": null, "release_date": "2018-10-15", "release_date_basis": "registry date", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Intel OpenVINO Open Model Zoo models for text detection, printed and handwritten text recognition and formula recognition.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/horizontal-text-detection-0001", "https://github.com/openvinotoolkit/open_model_zoo", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/formula-recognition-medium-scan-0001", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/handwritten-english-recognition-0001" ], "unverified": [] }, { "id": "text-recognition-resnet-fc-vitstr", "name": "Text recognition ResNet-FC and ViTSTR", "level": "model", "family_id": null, "developer": [], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Public scene text recognisers (ResNet with fully connected head, and the ViTSTR vision transformer) in the OpenVINO Open Model Zoo.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/text-recognition-resnet-fc" ], "unverified": [] }, { "id": "nvidia-ocdnet-ocrnet", "name": "OCDNet and OCRNet (TAO)", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "OCR and document", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "OCDNet text detection, OCRNet text recognition, nvCLIP4STR and MixNet scene text models.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA TAO optical character detection and recognition models for industrial OCR, published in the NGC catalog.", "verified_on": "2026-10-12", "sources": [ "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/ocdnet", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/nvclip4str", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/mixnet" ], "unverified": [] }, { "id": "cflow-ad", "name": "C-Flow (CFLOW-AD)", "level": "model", "family_id": null, "developer": [ "panasonic" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Anomalib groups it with CS-Flow (arXiv 2110.02855).", "context_length": null, "release_date": "2021-07-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Real-time unsupervised anomaly detection with conditional normalising flows.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2107.12571", "https://anomalib.readthedocs.io/en/latest/markdown/guides/reference/models/image/index.html", "https://arxiv.org/html/2107.12571" ], "unverified": [] }, { "id": "u-flow", "name": "U-Flow", "level": "model", "family_id": null, "developer": [ "digital-sense", "universidad-de-la-republica" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-11-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "U-shaped normalising flow for anomaly detection with an unsupervised threshold.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2211.12353", "https://anomalib.readthedocs.io/en/latest/markdown/guides/reference/models/image/index.html", "https://arxiv.org/html/2211.12353" ], "unverified": [ "developer" ] }, { "id": "inp-former", "name": "INP-Former", "level": "model", "family_id": null, "developer": [ "tsinghua-university", "hust" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Universal anomaly detection using intrinsic normal prototypes found within a single image.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2503.02424", "https://anomalib.readthedocs.io/en/latest/markdown/guides/reference/models/image/index.html", "https://arxiv.org/html/2503.02424" ], "unverified": [] }, { "id": "espcn", "name": "ESPCN (sub-pixel CNN)", "level": "model", "family_id": null, "developer": [ "twitter" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2016-09-16", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time image and video super-resolution with an efficient sub-pixel convolution layer; in the ONNX Model Zoo.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1609.05158", "https://github.com/onnx/models/tree/main/validated/vision/super_resolution/sub_pixel_cnn_2016", "https://arxiv.org/html/1609.05158" ], "unverified": [] }, { "id": "dncnn", "name": "DnCNN", "level": "model", "family_id": null, "developer": [ "harbin-institute-of-technology", "hong-kong-polytechnic-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2016-08-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Residual CNN for image denoising, distributed in the KAIR restoration toolbox.", "verified_on": "2026-10-12", "sources": [ "https://github.com/cszn/KAIR", "https://aihub.qualcomm.com/models/dncnn", "https://arxiv.org/html/1608.03981", "https://arxiv.org/abs/1608.03981" ], "unverified": [] }, { "id": "deblurgan", "name": "DeblurGAN", "level": "model", "family_id": null, "developer": [ "ukrainian-catholic-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-04-26", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "GAN for blind motion deblurring.", "verified_on": "2026-10-12", "sources": [ "https://github.com/KupynOrest/DeblurGAN", "https://modelzoo.co/model/deblurgan", "https://arxiv.org/html/1711.07064" ], "unverified": [] }, { "id": "google-maxim", "name": "MAXIM", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-01-09", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google multi-axis MLP for image processing: deblurring, dehazing, denoising, deraining and enhancement.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2201.02973", "https://replicate.com/google-research/maxim", "https://www.kaggle.com/models/kaggle/maxim", "https://github.com/tensorflow/tfhub.dev/tree/master/assets/docs/sayakpaul/collections/maxim/1.md" ], "unverified": [] }, { "id": "mirnet", "name": "MIRNet", "level": "model", "family_id": null, "developer": [ "inception-institute-of-ai" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multi-scale residual network for low-light image enhancement, published on Kaggle Models.", "verified_on": "2026-10-12", "sources": [ "https://www.kaggle.com/models/spsayakpaul/mirnet", "https://www.kaggle.com/models/kaggle/mirnet", "https://github.com/tensorflow/tfhub.dev/tree/master/assets/docs/rishit-dagli", "https://arxiv.org/html/2003.06792" ], "unverified": [] }, { "id": "gfpgan", "name": "GFPGAN", "level": "model", "family_id": null, "developer": [ "tencent" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-01-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Blind face restoration with a generative facial prior, from Tencent ARC.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2101.04061", "https://replicate.com/tencentarc/gfpgan", "https://arxiv.org/html/2101.04061" ], "unverified": [] }, { "id": "codeformer", "name": "CodeFormer", "level": "model", "family_id": null, "developer": [ "ntu-singapore" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-06-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Robust blind face restoration with a codebook lookup transformer, from NTU S-Lab.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2206.11253", "https://replicate.com/lucataco/codeformer", "https://arxiv.org/html/2206.11253" ], "unverified": [] }, { "id": "alibaba-damo-gpen", "name": "GPEN", "level": "model", "family_id": null, "developer": [ "alibaba-damo" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-05-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "GAN-prior embedded network for blind face restoration, published on ModelScope by Alibaba DAMO Academy.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2105.06070", "https://www.modelscope.cn/models/iic/cv_gpen_image-portrait-enhancement" ], "unverified": [] }, { "id": "deoldify", "name": "DeOldify", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open source", "licence": null, "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "GAN colouriser for old black-and-white photos and film by Jason Antic, trained with his NoGAN method; MIT licence, weights downloadable.", "verified_on": "2026-10-15", "sources": [ "https://github.com/jantic/DeOldify", "https://replicate.com/arielreplicate/deoldify_image" ] }, { "id": "supir", "name": "SUPIR", "level": "model", "family_id": null, "developer": [ "siat", "shanghai-ai-lab" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-01-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Generative photo-realistic image restoration and upscaling built on a large diffusion model.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2401.13627", "https://replicate.com/cjwbw/supir", "https://arxiv.org/html/2401.13627" ], "unverified": [] }, { "id": "rife", "name": "RIFE", "level": "model", "family_id": null, "developer": [ "megvii", "peking-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-11-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time intermediate flow estimation for video frame interpolation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2011.06294", "https://www.modelscope.cn/models/Damo_XR_Lab/cv_rife_video-frame-interpolation", "https://arxiv.org/html/2011.06294" ], "unverified": [] }, { "id": "qualcomm-quicksrnet", "name": "QuickSRNet", "level": "model", "family_id": null, "developer": [ "qualcomm" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qualcomm AI Research small super-resolution network for mobile, in the AIMET model zoo.", "verified_on": "2026-10-12", "sources": [ "https://github.com/quic/aimet-model-zoo/tree/develop/aimet_zoo_torch/quicksrnet", "https://aihub.qualcomm.com/models/quicksrnetsmall" ], "unverified": [] }, { "id": "xlsr-super-resolution", "name": "XLSR (super-resolution)", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Extremely lightweight quantisation-robust super-resolution network, in the Qualcomm AIMET model zoo.", "verified_on": "2026-10-12", "sources": [ "https://github.com/quic/aimet-model-zoo/tree/develop/aimet_zoo_torch/xlsr", "https://aihub.qualcomm.com/models/xlsr" ], "unverified": [] }, { "id": "intel-single-image-super-resolution", "name": "Intel single-image super-resolution", "level": "model", "family_id": null, "developer": [ "intel" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Intel OpenVINO super-resolution models (1032, 1033) that upscale images by 3 and 4 times.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/single-image-super-resolution-1032" ], "unverified": [] }, { "id": "fbcnn", "name": "FBCNN", "level": "model", "family_id": null, "developer": [ "eth-zurich" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Flexible blind JPEG artefact removal network, in the OpenVINO Open Model Zoo.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/fbcnn", "https://arxiv.org/html/2109.14573" ], "unverified": [] }, { "id": "sci-low-light", "name": "SCI low-light enhancement", "level": "model", "family_id": null, "developer": [ "dalian-university-of-technology" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: NXP eIQ model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Self-calibrated illumination network for low-light image enhancement; NXP ships an i.MX recipe.", "verified_on": "2026-10-12", "sources": [ "https://github.com/NXP/eiq-model-zoo/tree/main/tasks/vision/low-light-enhancement", "https://arxiv.org/html/2204.10137" ], "unverified": [] }, { "id": "mahmood-lab-conch", "name": "CONCH", "level": "model", "family_id": null, "developer": [ "mahmood-lab" ], "model_class": "vision", "modalities_in": [ "image", "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-01-05", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "gated preview" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Mahmood Lab vision-language foundation model for computational pathology, trained on histopathology image-caption pairs.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/MahmoodLab/CONCH", "https://huggingface.co/api/models?pipeline_tag=image-feature-extraction&sort=likes&limit=50" ], "unverified": [ "summary" ] }, { "id": "mahmood-lab-titan", "name": "TITAN", "level": "model", "family_id": null, "developer": [ "mahmood-lab" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": 158866176, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-11-29", "release_date_basis": "announcement page", "weights_date": "2024-12-02", "openness": "open weights restricted", "licence": null, "access_routes": [ "gated preview" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Mahmood Lab multimodal whole-slide foundation model for pathology.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2411.19666", "https://huggingface.co/MahmoodLab/TITAN" ], "unverified": [] }, { "id": "mahmood-lab-uni2-h", "name": "UNI2-h", "level": "model", "family_id": null, "developer": [ "mahmood-lab" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-01-10", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "gated preview" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Second-generation UNI pathology foundation encoder from Mahmood Lab.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/MahmoodLab/UNI2-h", "https://huggingface.co/api/models?pipeline_tag=image-feature-extraction&sort=likes&limit=50" ], "unverified": [ "summary" ] }, { "id": "paige-virchow2", "name": "Virchow2", "level": "model", "family_id": null, "developer": [ "paige" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": 631239424, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-08-01", "release_date_basis": "announcement page", "weights_date": "2024-08-05", "openness": "open weights restricted", "licence": null, "access_routes": [ "gated preview" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Paige pathology foundation model trained with mixed magnifications; successor of Virchow.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2408.00738", "https://huggingface.co/paige-ai/Virchow2", "https://huggingface.co/api/models?pipeline_tag=image-feature-extraction&sort=likes&limit=50" ], "unverified": [] }, { "id": "h-optimus-0", "name": "H-optimus-0", "level": "model", "family_id": null, "developer": [ "bioptimus" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-07-30", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "cloud marketplace", "gated preview" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Pathology foundation vision transformer published by Bioptimus on Hugging Face.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/bioptimus/H-optimus-0", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://huggingface.co/api/models?pipeline_tag=image-feature-extraction&sort=likes&limit=50" ], "unverified": [] }, { "id": "hibou", "name": "Hibou", "level": "model", "family_id": null, "developer": [ "histai" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 85741056, "params_active": null, "params_checkpoint": null, "params_note": "Hibou-B and Hibou-L.", "context_length": null, "release_date": "2024-06-07", "release_date_basis": "announcement page", "weights_date": "2024-06-06", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "cloud marketplace", "gated preview" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Family of pathology foundation vision transformers from HistAI.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2406.05074", "https://huggingface.co/histai/hibou-b", "https://ai.azure.com/catalog/models?industry=health-and-life-sciences", "https://arxiv.org/html/2406.05074" ], "unverified": [] }, { "id": "microsoft-rad-dino", "name": "RAD-DINO", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": 86580480, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-05-17", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft chest X-ray image encoder trained with DINOv2 self-supervision, for radiology classification, segmentation and report generation.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/microsoft/rad-dino", "https://ai.azure.com/catalog/models?industry=health-and-life-sciences" ], "unverified": [] }, { "id": "google-cxr-foundation", "name": "CXR Foundation", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-11-20", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "gated preview" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google chest X-ray embedding model for building classifiers with little labelled data (Health AI Developer Foundations).", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/google/cxr-foundation", "https://huggingface.co/api/models?pipeline_tag=image-classification&sort=likes&limit=50", "https://huggingface.co/google/derm-foundation" ], "unverified": [ "summary" ] }, { "id": "google-derm-foundation", "name": "Derm Foundation", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-11-20", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "gated preview" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google dermatology image embedding model for skin condition classifiers (Health AI Developer Foundations).", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/google/derm-foundation", "https://huggingface.co/api/models?pipeline_tag=image-classification&sort=likes&limit=50" ], "unverified": [ "summary" ] }, { "id": "microsoft-medimageparse", "name": "MedImageParse (BiomedParse)", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "MedImageParse 3D is offered in Azure AI Foundry.", "context_length": null, "release_date": "2024-11-04", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-sa-4-0", "access_routes": [ "cloud marketplace", "gated preview" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft biomedical model for joint segmentation, detection and recognition across nine imaging modalities from text prompts.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/microsoft/BiomedParse", "https://learn.microsoft.com/en-us/azure/foundry-classic/how-to/healthcare-ai/healthcare-ai-models" ], "unverified": [ "summary" ] }, { "id": "gigatime", "name": "GigaTIME", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10-21", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "cloud marketplace", "gated preview" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Pathology model that translates H&E slides into virtual multiplex immunofluorescence, published by the prov-gigatime organisation on Hugging Face.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/prov-gigatime/GigaTIME", "https://ai.azure.com/catalog/models?industry=health-and-life-sciences" ], "unverified": [ "summary" ] }, { "id": "stanford-university-medvae", "name": "MedVAE", "level": "model", "family_id": null, "developer": [ "stanford-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-02-20", "release_date_basis": "announcement page", "weights_date": "2025-01-29", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Stanford autoencoders that compress medical images into small latents for efficient downstream interpretation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2502.14753", "https://huggingface.co/stanfordmimi/MedVAE", "https://ai.azure.com/catalog/models?industry=health-and-life-sciences" ], "unverified": [] }, { "id": "nvidia-monai-bundles", "name": "MONAI model bundles", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "BraTS MRI, spleen CT, pancreas CT, whole-body CT and Swin UNETR bundles.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MONAI consortium medical imaging segmentation bundles built and published by NVIDIA in the NGC catalog.", "verified_on": "2026-10-12", "sources": [ "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/monaitoolkit/models/monai_brats_mri_segmentation" ], "unverified": [] }, { "id": "nvidia-vista", "name": "VISTA-2D and VISTA-3D", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA foundation models for cell segmentation (VISTA-2D) and interactive 3D CT segmentation (VISTA-3D).", "verified_on": "2026-10-12", "sources": [ "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/monaitoolkit/models/monai_vista2d" ], "unverified": [] }, { "id": "nvidia-polyp-detection-rt-detr", "name": "Polyp Detection RT-DETR (Holoscan)", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA RT-DETR model for real-time polyp detection in endoscopy video on the Holoscan platform.", "verified_on": "2026-10-12", "sources": [ "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/clara-holoscan/models/polyp_detection_rt_detr", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/clara-holoscan/models/polyp_detection_rt_detr_model" ], "unverified": [] }, { "id": "act-action-chunking-transformer", "name": "ACT (Action Chunking with Transformers)", "level": "model", "family_id": null, "developer": [ "stanford-university" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-04-23", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU", "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Imitation learning policy that predicts chunks of robot actions with a transformer, introduced with the low-cost ALOHA bimanual arms; LeRobot, OpenVINO and Qualcomm AI Hub ship it.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2304.13705", "https://github.com/tonyzhaozh/act", "https://aihub.qualcomm.com/models/act", "https://huggingface.co/OpenVINO/act-fp16-ov", "https://www.modelscope.cn/models/lerobot/act_aloha_sim_transfer_cube_human", "https://github.com/open-mmlab/mmaction2/tree/main/projects/actionclip", "https://arxiv.org/abs/2109.08472", "https://arxiv.org/html/2304.13705" ], "unverified": [] }, { "id": "diffusion-policy", "name": "Diffusion Policy", "level": "model", "family_id": null, "developer": [ "columbia-university" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": "diffusion", "architecture_detail": null, "params_total": 262709062, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-03-07", "release_date_basis": "announcement page", "weights_date": "2024-05-05", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Visuomotor robot policy that generates actions by diffusion; LeRobot publishes a PushT checkpoint.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2303.04137", "https://huggingface.co/lerobot/diffusion_pusht", "https://huggingface.co/api/models?pipeline_tag=robotics&sort=downloads&limit=50", "https://www.modelscope.cn/models/lerobot/diffusion_pusht", "https://arxiv.org/html/2303.04137" ], "unverified": [] }, { "id": "hugging-face-smolvla", "name": "SmolVLA", "level": "model", "family_id": null, "developer": [ "hugging-face" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-06-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Hugging Face LeRobot small vision-language-action model for affordable robots; OpenVINO publishes a LIBERO build.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2506.01844", "https://www.modelscope.cn/models/lerobot/smolvla_base", "https://huggingface.co/OpenVINO/smolvla-libero-fp16-ov" ], "unverified": [] }, { "id": "x-vla", "name": "X-VLA", "level": "model", "family_id": null, "developer": [ "tsinghua-university", "shanghai-ai-lab" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 879738545, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10-11", "release_date_basis": "announcement page", "weights_date": "2025-11-19", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Soft-prompted transformer vision-language-action model for cross-embodiment robot control, ported to LeRobot.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2510.10274", "https://huggingface.co/lerobot/xvla-base", "https://huggingface.co/api/models?pipeline_tag=robotics&sort=downloads&limit=50", "https://arxiv.org/html/2510.10274" ], "unverified": [] }, { "id": "rdt-1b", "name": "RDT-1B", "level": "model", "family_id": null, "developer": [ "tsinghua-university" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": "diffusion", "architecture_detail": null, "params_total": 1000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-10-10", "release_date_basis": "announcement page", "weights_date": "2024-08-27", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Diffusion foundation model with about 1B parameters for bimanual robot manipulation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2410.07864", "https://huggingface.co/robotics-diffusion-transformer/rdt-1b", "https://huggingface.co/api/models?pipeline_tag=robotics&sort=likes&limit=50", "https://arxiv.org/html/2410.07864" ], "unverified": [] }, { "id": "baai-robobrain-2-0", "name": "RoboBrain 2.0", "level": "model", "family_id": null, "developer": [ "baai" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8292166656, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-07-02", "release_date_basis": "announcement page", "weights_date": "2025-06-05", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "gated preview" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BAAI embodied vision-language model for robot planning, spatial reasoning and trajectory prediction.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2507.02029", "https://huggingface.co/BAAI/RoboBrain2.0-7B", "https://huggingface.co/api/models?pipeline_tag=robotics&sort=likes&limit=50", "https://www.modelscope.cn/models/BAAI/RoboBrain2.0-7B" ], "unverified": [] }, { "id": "allen-ai-molmoact2", "name": "MolmoAct2", "level": "model", "family_id": null, "developer": [ "allen-ai" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 5442196272, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-05-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Ai2 action reasoning model for real-world robot deployment, built on Molmo.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2605.02881", "https://huggingface.co/allenai/MolmoAct2", "https://huggingface.co/api/models?pipeline_tag=robotics&sort=downloads&limit=50" ], "unverified": [] }, { "id": "openbmb-minicpm-robot", "name": "MiniCPM RobotManip and RobotTrack", "level": "model", "family_id": null, "developer": [ "openbmb" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 1536237296, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-07-18", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenBMB MiniCPM models for robot manipulation and target tracking, published on Hugging Face.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/openbmb/MiniCPM-RobotManip", "https://huggingface.co/api/models?pipeline_tag=robotics&sort=likes&limit=50" ], "unverified": [] }, { "id": "dobb-e", "name": "Dobb-E", "level": "model", "family_id": null, "developer": [ "nyu", "meta" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": "CNN", "architecture_detail": null, "params_total": 21301696, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-11-27", "release_date_basis": "announcement page", "weights_date": "2023-11-24", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Home robot learning framework and pretrained visual representation from New York University, from the paper \"On Bringing Robots Home\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2311.16098", "https://huggingface.co/notmahi/dobb-e", "https://huggingface.co/api/models?pipeline_tag=robotics&sort=likes&limit=50", "https://arxiv.org/html/2311.16098" ], "unverified": [] }, { "id": "black-forest-labs-flux-3-action-base", "name": "FLUX.3 Action Base", "level": "model", "family_id": null, "developer": [ "black-forest-labs" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": "flow matching", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-09-22", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Black Forest Labs base model for robot actions in the FLUX.3 line, published on Hugging Face.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/black-forest-labs/flux-3-action-base", "https://huggingface.co/api/models?pipeline_tag=robotics&sort=likes&limit=50" ], "unverified": [] }, { "id": "nvidia-gear-sonic", "name": "GEAR-SONIC", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-11-11", "release_date_basis": "announcement page", "weights_date": "2026-02-12", "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA GEAR lab humanoid whole-body control policy trained by large-scale motion tracking.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2511.07820", "https://huggingface.co/nvidia/GEAR-SONIC", "https://huggingface.co/api/models?pipeline_tag=reinforcement-learning&sort=downloads&limit=50" ], "unverified": [] }, { "id": "nvidia-nitrogen", "name": "NitroGen", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "robot policy", "modalities_in": [ "image", "text" ], "modalities_out": [ "action" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01-04", "release_date_basis": "announcement page", "weights_date": "2025-12-18", "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA open foundation model for generalist gaming agents that maps game video to controller actions.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2601.02427", "https://huggingface.co/nvidia/NitroGen", "https://huggingface.co/api/models?pipeline_tag=reinforcement-learning&sort=likes&limit=50" ], "unverified": [] }, { "id": "meta-sparsh", "name": "Sparsh", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "robot policy", "modalities_in": [ "sensor" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta self-supervised touch representation models for vision-based tactile sensors, published on Kaggle.", "verified_on": "2026-10-12", "sources": [ "https://www.kaggle.com/models/metaresearch/sparsh" ], "unverified": [ "modalities_out" ] }, { "id": "nvidia-foundationpose", "name": "FoundationPose", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "robot policy", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA 6D object pose estimation and tracking model for novel objects, used in Isaac ROS robot manipulation.", "verified_on": "2026-10-12", "sources": [ "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/isaac/models/foundationpose" ], "unverified": [ "modalities_out" ] }, { "id": "nvidia-centerpose", "name": "CenterPose", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA TAO and Isaac ROS category-level 3D object pose estimation model.", "verified_on": "2026-10-12", "sources": [ "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/centerpose" ], "unverified": [] }, { "id": "bevformer", "name": "BEVFormer", "level": "model", "family_id": null, "developer": [ "nanjing-university", "shanghai-ai-lab" ], "model_class": "driving", "modalities_in": [ "image", "sensor" ], "modalities_out": [ "3D" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-03-31", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub, TI edge AI model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Bird's-eye-view perception from multiple cameras with spatiotemporal transformers, for 3D detection and map segmentation in autonomous driving.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2203.17270", "https://github.com/fundamentalvision/BEVFormer", "https://aihub.qualcomm.com/models/bevformer", "https://github.com/fundamentalvision/BEVFormer/", "https://github.com/TexasInstruments/edgeai-tensorlab/tree/main/edgeai-modelzoo/models/vision/detection_3d/pandaset", "https://arxiv.org/html/2203.17270" ], "unverified": [] }, { "id": "bevdet", "name": "BEVDet", "level": "model", "family_id": null, "developer": [ "phigent-robotics" ], "model_class": "driving", "modalities_in": [ "image", "sensor" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-12-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multi-camera 3D object detection in bird's-eye view, from PhiGent Robotics.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2112.11790", "https://github.com/HuangJunJie2017/BEVDet", "https://aihub.qualcomm.com/models/bevdet", "https://github.com/HuangJunJie2017/BEVDet/", "https://arxiv.org/html/2112.11790" ], "unverified": [] }, { "id": "cross-view-transformers", "name": "CVT (Cross-View Transformers)", "level": "model", "family_id": null, "developer": [ "ut-austin" ], "model_class": "driving", "modalities_in": [ "image", "sensor" ], "modalities_out": [ "3D" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-05-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time map-view semantic segmentation from multiple cameras.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2205.02833", "https://github.com/bradyz/cross_view_transformers", "https://aihub.qualcomm.com/models/cvt", "https://arxiv.org/html/2205.02833" ], "unverified": [] }, { "id": "hust-gkt", "name": "GKT", "level": "model", "family_id": null, "developer": [ "hust" ], "model_class": "driving", "modalities_in": [ "image", "sensor" ], "modalities_out": [ "3D" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-06-09", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Geometry-guided kernel transformer for efficient 2D-to-BEV representation learning.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2206.04584", "https://github.com/hustvl/GKT", "https://aihub.qualcomm.com/models/gkt", "https://github.com/hustvl/GKT/" ], "unverified": [] }, { "id": "simple-bev", "name": "Simple-BEV", "level": "model", "family_id": null, "developer": [ "stanford-university", "carnegie-mellon-university" ], "model_class": "driving", "modalities_in": [ "image", "sensor" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-06-16", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Simple multi-sensor bird's-eye-view perception baseline.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2206.07959", "https://github.com/aharley/simple_bev", "https://aihub.qualcomm.com/models/simple_bev_cam", "https://github.com/aharley/simple_bev/blob/main/nets/segnet.py", "https://arxiv.org/html/2206.07959" ], "unverified": [] }, { "id": "salsanext", "name": "SalsaNext", "level": "model", "family_id": null, "developer": [ "halmstad-university", "volvo" ], "model_class": "driving", "modalities_in": [ "image", "sensor" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-03-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fast uncertainty-aware semantic segmentation of LiDAR point clouds for autonomous driving.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2003.03653", "https://aihub.qualcomm.com/models/salsanext", "https://arxiv.org/html/2003.03653" ], "unverified": [] }, { "id": "rangenet-plus-plus", "name": "RangeNet++", "level": "model", "family_id": null, "developer": [], "model_class": "driving", "modalities_in": [ "image", "sensor" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-07-24", "release_date_basis": "registry date", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "LiDAR semantic segmentation on range images, from the University of Bonn.", "verified_on": "2026-10-12", "sources": [ "https://github.com/PRBonn/lidar-bonnetal", "https://aihub.qualcomm.com/models/rangenet_plus_plus" ], "unverified": [] }, { "id": "statetransformer", "name": "StateTransformer", "level": "model", "family_id": null, "developer": [], "model_class": "driving", "modalities_in": [ "sensor" ], "modalities_out": [ "action" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Autoregressive transformer for autonomous driving motion planning; Qualcomm AI Hub ships a device build.", "verified_on": "2026-10-12", "sources": [ "https://aihub.qualcomm.com/models/statetransformer" ], "unverified": [] }, { "id": "3d-deep-box", "name": "3D-Deep-Box", "level": "model", "family_id": null, "developer": [ "george-mason-university", "zoox" ], "model_class": "driving", "modalities_in": [ "image", "sensor" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "3D bounding box estimation from monocular images (Mousavian et al.), with a PyTorch implementation used by Qualcomm AI Hub.", "verified_on": "2026-10-12", "sources": [ "https://github.com/skhadem/3D-BoundingBox", "https://aihub.qualcomm.com/models/deepbox", "https://github.com/skhadem/3D-BoundingBox/", "https://arxiv.org/html/1612.00496" ], "unverified": [] }, { "id": "lanenet", "name": "LaneNet", "level": "model", "family_id": null, "developer": [ "ku-leuven" ], "model_class": "driving", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-02-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "End-to-end lane detection as instance segmentation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1802.05591", "https://github.com/MaybeShewill-CV/lanenet-lane-detection", "https://modelzoo.co/model/lanenet-lane-detection", "https://arxiv.org/html/1802.05591" ], "unverified": [] }, { "id": "hust-yolop", "name": "YOLOP", "level": "model", "family_id": null, "developer": [ "hust" ], "model_class": "driving", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "YOLOP and YOLOPv2 (arXiv 2208.11434).", "context_length": null, "release_date": "2021-08-25", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Panoptic driving perception network that detects vehicles, drivable area and lanes in one pass.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2108.11250", "https://github.com/hustvl/YOLOP", "https://www.modelscope.cn/models/iic/cv_yolopv2_image-driving-perception_bdd100k", "https://pytorch.org/hub/hustvl_yolop", "https://raw.githubusercontent.com/pytorch/hub/master/hustvl_yolop.md" ], "unverified": [] }, { "id": "hybridnets", "name": "HybridNets", "level": "model", "family_id": null, "developer": [], "model_class": "driving", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-03-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "End-to-end perception network for traffic object detection, drivable area and lane segmentation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2203.09035", "https://github.com/datvuthanh/HybridNets", "https://pytorch.org/hub/datvuthanh_hybridnets", "https://raw.githubusercontent.com/pytorch/hub/master/datvuthanh_hybridnets.md" ], "unverified": [] }, { "id": "intel-road-segmentation-adas", "name": "Intel road and ADAS segmentation", "level": "model", "family_id": null, "developer": [ "intel" ], "model_class": "driving", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Intel OpenVINO models that segment road, curb and markings for driver assistance.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/road-segmentation-adas-0001" ], "unverified": [] }, { "id": "dkaze-visual-localisation", "name": "DKAZE visual localisation (TI)", "level": "model", "family_id": null, "developer": [], "model_class": "driving", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: TI edge AI model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Texas Instruments deep KAZE feature model for visual localisation, trained on CARLA simulation and shipped in the TI edge AI model zoo.", "verified_on": "2026-10-12", "sources": [ "https://github.com/TexasInstruments/edgeai-tensorlab/tree/main/edgeai-modelzoo/models/vision", "https://github.com/TexasInstruments/edgeai-tensorlab/tree/main/edgeai-modelzoo/models/vision/visual_localization/carla" ], "unverified": [] }, { "id": "oriented-object-detectors", "name": "Oriented object detectors for aerial images (MMRotate)", "level": "family", "family_id": null, "developer": [ "openmmlab" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Also in MMRotate: Rotated RetinaNet, Rotated Faster R-CNN, Rotated FCOS, Rotated RepPoints, Rotated ATSS, CFA, Oriented RepPoints (arXiv 2105.11111), SASM, G-Rep; losses CSL, GWD, KLD and KFIoU.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Detectors that predict rotated boxes for ships, planes, vehicles and storage tanks in aerial and satellite images, reproduced with checkpoints on DOTA in OpenMMLab MMRotate.", "verified_on": "2026-10-12", "sources": [ "https://github.com/open-mmlab/mmrotate", "https://github.com/open-mmlab/mmrotate/tree/main/configs/rotated_retinanet", "https://arxiv.org/abs/1708.02002", "https://github.com/open-mmlab/mmrotate/tree/main/configs/rotated_faster_rcnn", "https://arxiv.org/abs/1506.01497", "https://github.com/open-mmlab/mmrotate/tree/main/configs/rotated_reppoints", "https://arxiv.org/abs/1904.11490" ], "unverified": [] }, { "id": "roi-transformer", "name": "RoI Transformer", "level": "variant", "family_id": "oriented-object-detectors", "developer": [ "wuhan-university" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-12-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Detector that learns rotated regions of interest for oriented objects in aerial images.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1812.00155", "https://arxiv.org/html/1812.00155" ], "unverified": [] }, { "id": "oriented-r-cnn", "name": "Oriented R-CNN", "level": "variant", "family_id": "oriented-object-detectors", "developer": [ "nwpu" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-08-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Two-stage oriented object detector with an oriented region proposal network.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2108.05699", "https://arxiv.org/html/2108.05699" ], "unverified": [] }, { "id": "redet", "name": "ReDet", "level": "variant", "family_id": "oriented-object-detectors", "developer": [ "wuhan-university" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-03-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Rotation-equivariant detector for aerial object detection.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.07733", "https://arxiv.org/html/2103.07733" ], "unverified": [] }, { "id": "s2a-net", "name": "S2A-Net", "level": "variant", "family_id": "oriented-object-detectors", "developer": [ "wuhan-university" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-08-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Single-shot alignment network for oriented object detection, from the paper \"Align Deep Features for Oriented Object Detection\".", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2008.09397", "https://arxiv.org/html/2008.09397" ], "unverified": [] }, { "id": "r3det", "name": "R3Det", "level": "variant", "family_id": "oriented-object-detectors", "developer": [ "shanghai-jiao-tong-university" ], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-08-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Refined single-stage detector for rotated objects.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1908.05612", "https://arxiv.org/html/1908.05612" ], "unverified": [] }, { "id": "gliding-vertex", "name": "Gliding Vertex", "level": "variant", "family_id": "oriented-object-detectors", "developer": [], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-11-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Multi-oriented object detection by gliding the vertices of horizontal boxes.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1911.09358" ], "unverified": [] }, { "id": "intel-person-detection", "name": "Intel person detection models", "level": "model", "family_id": null, "developer": [ "intel" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "person-detection 0106, 0200 to 0203, 0301 to 0303, retail; pedestrian-and-vehicle and person-vehicle-bike detectors.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Intel OpenVINO detectors for people, pedestrians, vehicles and bikes in retail, security and traffic cameras.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/person-detection-0200", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/pedestrian-and-vehicle-detector-adas-0001", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/person-vehicle-bike-detection-2000", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/person-detection-0106" ], "unverified": [] }, { "id": "intel-face-detection", "name": "Intel face models", "level": "model", "family_id": null, "developer": [ "intel" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "face-detection 0200 to 0206 and ADAS, face-reidentification-retail-0095, facial landmarks 35 and 98, age-gender-recognition-retail-0013, emotions-recognition-retail-0003, gaze and head pose estimation.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Intel OpenVINO face detection, landmark, re-identification, age, gender, emotion and gaze models for retail analytics and driver monitoring.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/face-detection-0200", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/age-gender-recognition-retail-0013", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/emotions-recognition-retail-0003", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/face-reidentification-retail-0095", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/facial-landmarks-35-adas-0002", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/gaze-estimation-adas-0002" ], "unverified": [] }, { "id": "intel-person-attributes-reid", "name": "Intel person attributes and re-identification", "level": "model", "family_id": null, "developer": [ "intel" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "person-reidentification-retail 0277, 0286, 0287, 0288; person-attributes-recognition-crossroad 0230, 0234, 0238; vehicle attributes.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Intel OpenVINO models for person re-identification and person and vehicle attribute recognition across cameras.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/person-reidentification-retail-0277", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/person-attributes-recognition-crossroad-0230" ], "unverified": [] }, { "id": "intel-licence-plate-recognition", "name": "Intel licence plate recognition", "level": "model", "family_id": null, "developer": [ "intel" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "license-plate-recognition-barrier 0001 and 0007, vehicle-license-plate-detection-barrier.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Intel OpenVINO models that detect vehicles and licence plates and read Chinese plates at barriers.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/license-plate-recognition-barrier-0001", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/license-plate-recognition-barrier-0007" ], "unverified": [] }, { "id": "intel-human-pose-estimation", "name": "Intel human pose estimation", "level": "model", "family_id": null, "developer": [ "intel" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "human-pose-estimation 0001, 0005, 0006, 0007; public 3D and single-person models.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Intel OpenVINO multi-person 2D pose estimation models (OpenPose-style and higher-resolution versions).", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/human-pose-estimation-0001", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/human-pose-estimation-3d-0001" ], "unverified": [] }, { "id": "intel-action-recognition", "name": "Intel action recognition", "level": "model", "family_id": null, "developer": [ "intel" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "action-recognition-0001, driver-action-recognition, classroom person-detection-action-recognition, asl-recognition, Smart Lab models, YOLOv2 AVA.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Intel OpenVINO models for action, driver activity, classroom behaviour and sign language recognition.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/action-recognition-0001", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/asl-recognition-0004", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/person-detection-action-recognition-0005", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/smartlab-action-recognition-0001", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/yolo-v2-ava-0001" ], "unverified": [] }, { "id": "intel-instance-segmentation", "name": "Intel instance and semantic segmentation", "level": "model", "family_id": null, "developer": [ "intel" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "instance-segmentation-person-0007, security 0002, 0091, 0228, 1039, 1040; unet-camvid-onnx-0001; ICNet CamVid.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Intel OpenVINO instance and semantic segmentation models for people, security scenes and road scenes.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/instance-segmentation-person-0007", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/icnet-camvid-ava-0001", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/unet-camvid-onnx-0001" ], "unverified": [] }, { "id": "intel-weld-porosity-detection", "name": "Intel weld porosity detection", "level": "model", "family_id": null, "developer": [ "intel" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Intel OpenVINO classifier that detects porosity defects in welding video for industrial inspection.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/weld-porosity-detection-0001" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "intel-product-detection", "name": "Intel product detection", "level": "model", "family_id": null, "developer": [ "intel" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Intel OpenVINO detector for retail products on shelves and at checkout.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/product-detection-0001" ], "unverified": [] }, { "id": "intel-other-omz-models", "name": "Intel binary, sparse and retrieval models", "level": "model", "family_id": null, "developer": [ "intel" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "resnet18 and resnet50 xnor binary, faster-rcnn-resnet101-coco-sparse-60-0001, image-retrieval-0001.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Intel OpenVINO binary-weight ResNets, a sparse Faster R-CNN and an image retrieval model showing compression and retrieval on CPUs.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/image-retrieval-0001", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/faster-rcnn-resnet101-coco-sparse-60-0001", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/resnet18-xnor-binary-onnx-0001" ], "unverified": [] }, { "id": "nvidia-tao-detectnet", "name": "NVIDIA TAO pretrained detection and segmentation models", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "DetectNet_v2 and other TAO pretrained backbones, FAN, FasterViT and GCViT backbones, PeopleSegNet and PeopleSemSegNet, Mask Auto Label, SegIC, SyntheticaDETR, RT-DETR 2D Warehouse.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA TAO Toolkit pretrained detection, segmentation and backbone models for fine-tuning and deployment with DeepStream and Isaac.", "verified_on": "2026-10-12", "sources": [ "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/pretrained_classification", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/pretrained_fan_classification_imagenet", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/mask_auto_label", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/segic", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/isaac/models/synthetica_detr", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/rtdetr_2d_warehouse", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/isaac/models/optimized-peoplesemseg-amr" ], "unverified": [] }, { "id": "nvidia-dashcamnet", "name": "DashCamNet", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA TAO detector for cars, people, road signs and two-wheelers seen from a moving dashboard camera.", "verified_on": "2026-10-12", "sources": [ "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/dashcamnet" ], "unverified": [] }, { "id": "nvidia-lpdnet-lprnet", "name": "LPDNet and LPRNet", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA TAO licence plate detection and recognition models for parking and traffic systems.", "verified_on": "2026-10-12", "sources": [ "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/lpdnet" ], "unverified": [] }, { "id": "nvidia-vehiclemakenet", "name": "VehicleMakeNet and VehicleTypeNet", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA TAO classifiers for vehicle make and vehicle type in traffic cameras.", "verified_on": "2026-10-12", "sources": [ "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/vehiclemakenet" ], "unverified": [] }, { "id": "nvidia-facedetect", "name": "FaceDetect and FaceDetectIR", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA TAO face detectors for RGB and infrared cameras.", "verified_on": "2026-10-12", "sources": [ "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/facedetectir" ], "unverified": [] }, { "id": "nvidia-tao-human-models", "name": "NVIDIA TAO human understanding models", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "BodyPoseNet, BodyPose3DNet, pose classification, ActionRecognitionNet, EmotionNet, GazeNet, GestureNet, HeartRateNet, facial landmarks.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA TAO models for body pose, actions, emotion, gaze, gestures and heart rate from video.", "verified_on": "2026-10-12", "sources": [ "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/bodyposenet", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/emotionnet", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/actionrecognitionnet" ], "unverified": [ "summary" ] }, { "id": "nvidia-reidentificationnet", "name": "ReIdentificationNet", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA TAO person re-identification models (CNN and transformer) for multi-camera tracking.", "verified_on": "2026-10-12", "sources": [ "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/reidentificationnet" ], "unverified": [] }, { "id": "nvidia-retail-object-recognition", "name": "Retail Object Detection and Recognition", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA models that detect and recognise retail items for checkout and shelf applications.", "verified_on": "2026-10-12", "sources": [ "https://catalog.ngc.nvidia.com/orgs/nvaie/models/retail_object_recognition" ], "unverified": [] }, { "id": "nvidia-pcb-defect-classification", "name": "Optical Inspection and PCB Defect Classification", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA models for automated optical inspection that classify printed circuit board defects.", "verified_on": "2026-10-12", "sources": [ "https://catalog.ngc.nvidia.com/orgs/nvaie/models/pcbclassification" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "nvidia-visual-changenet", "name": "Visual ChangeNet", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA TAO change detection models (classification and segmentation) that compare two images for defects or scene changes.", "verified_on": "2026-10-12", "sources": [ "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/visual_changenet_changesim" ], "unverified": [ "summary", "indonesian_measured", "indonesia_hosting" ] }, { "id": "nvidia-c-radio", "name": "C-RADIOv2", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "RADIO-CLIP is the related image-text model.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA agglomerative vision foundation encoder distilled from several teacher models, published in the TAO catalog.", "verified_on": "2026-10-12", "sources": [ "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/cradiov2" ], "unverified": [] }, { "id": "nvidia-nv-dinov2", "name": "NV-DINOv2", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA build of Meta DINOv2 with a ViT-L backbone, for image classification and as a TAO backbone.", "verified_on": "2026-10-12", "sources": [ "https://catalog.ngc.nvidia.com/orgs/nvaie/models/imagenet_nv_dinov2" ], "unverified": [] }, { "id": "nvidia-latent-landmarks", "name": "Unsupervised Latent Landmarks", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA model that discovers object landmarks without labels, published in the NGC catalog.", "verified_on": "2026-10-12", "sources": [ "https://catalog.ngc.nvidia.com/orgs/nvidia/models/latent_landmarks_bbc_fgbg_fixedcovariance" ], "unverified": [] }, { "id": "qualcomm-gearguardnet", "name": "PPE-Detection (GearGuardNet)", "level": "model", "family_id": null, "developer": [ "qualcomm" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qualcomm object detector that checks in real time whether a person is wearing personal protective equipment (PPE).", "verified_on": "2026-10-12", "sources": [ "https://aihub.qualcomm.com/models/gear_guard_net" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "qualcomm-face-models", "name": "Qualcomm face models", "level": "model", "family_id": null, "developer": [ "qualcomm" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Lightweight-Face-Detection, Facial-Attribute-Detection, Facial-Landmark-Detection (FaceMap 3DMM).", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qualcomm on-device models for face detection, facial attributes and 3D facial landmarks.", "verified_on": "2026-10-12", "sources": [ "https://aihub.qualcomm.com/models/face_det_lite", "https://aihub.qualcomm.com/models/face_attrib_net", "https://aihub.qualcomm.com/models/facemap_3dmm" ], "unverified": [] }, { "id": "qualcomm-person-foot-detection", "name": "Person-Foot-Detection", "level": "model", "family_id": null, "developer": [ "qualcomm" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qualcomm detector for people and their feet for floor-position tracking, in the AI Hub models repository.", "verified_on": "2026-10-12", "sources": [ "https://github.com/qualcomm/ai-hub-models", "https://aihub.qualcomm.com/models/foot_track_net", "https://github.com/qualcomm/ai-hub-models/blob/main/src/qai_hub_models/models/foot_track_net/model.py" ], "unverified": [] }, { "id": "bgnet", "name": "BGNet", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03-13", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Segmentation model shipped by Qualcomm AI Hub for on-device use.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/qualcomm/BGNet" ], "unverified": [ "summary" ] }, { "id": "eyegaze", "name": "EyeGaze", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Eye gaze estimation network; Qualcomm AI Hub ships a device build.", "verified_on": "2026-10-12", "sources": [ "https://github.com/david-wb/gaze-estimation", "https://aihub.qualcomm.com/models/eyegaze" ], "unverified": [] }, { "id": "google-speciesnet", "name": "SpeciesNet", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google camera-trap model that detects animals and classifies their species, for wildlife monitoring.", "verified_on": "2026-10-12", "sources": [ "https://www.kaggle.com/models/google/speciesnet" ], "unverified": [] }, { "id": "google-circularnet", "name": "CircularNet", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google waste identification models that segment and classify materials on recycling conveyor belts.", "verified_on": "2026-10-12", "sources": [ "https://www.kaggle.com/models/google/circularnet", "https://github.com/tensorflow/tfhub.dev/tree/master/assets/docs/google", "https://github.com/tensorflow/models/tree/master/official/projects/waste_identification_ml" ], "unverified": [] }, { "id": "google-cropnet", "name": "CropNet", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google cassava leaf disease classifier and plant feature vector models.", "verified_on": "2026-10-12", "sources": [ "https://www.kaggle.com/models/google/cropnet", "https://github.com/tensorflow/tfhub.dev/tree/master/assets/docs/google" ], "unverified": [ "summary" ] }, { "id": "google-cots-detector", "name": "COTS detector", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google detector for crown-of-thorns starfish in underwater video, for reef protection; in the TensorFlow Model Garden.", "verified_on": "2026-10-12", "sources": [ "https://github.com/tensorflow/models/tree/master/official/projects/cots_detector" ], "unverified": [] }, { "id": "google-musiq", "name": "MUSIQ", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Kaggle groups it with VILA aesthetics models.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google multi-scale image quality transformer that scores photo quality and aesthetics.", "verified_on": "2026-10-12", "sources": [ "https://www.kaggle.com/models/google/musiq", "https://www.kaggle.com/models/google/vila" ], "unverified": [] }, { "id": "google-aiy-vision-classifiers", "name": "AIY food, product and nature classifiers", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "AIY food classifier, Mobile Food Segmenter, popular US products, popular wine and fungi classifiers.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google mobile classifiers for food dishes, packaged products, wine labels and fungi, published on Kaggle.", "verified_on": "2026-10-12", "sources": [ "https://www.kaggle.com/models/google/aiy", "https://www.kaggle.com/models/google/mobile-food-segmenter-v1", "https://www.kaggle.com/models/google/popular-us-products-v1" ], "unverified": [ "summary" ] }, { "id": "google-mobile-object-localizer", "name": "Mobile Object Localizer", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google class-agnostic mobile object localiser and labeller, published on Kaggle.", "verified_on": "2026-10-12", "sources": [ "https://www.kaggle.com/models/google/mobile-object-localizer-v1", "https://www.kaggle.com/models/google/mobile-object-labeler-v1" ], "unverified": [] }, { "id": "microsoft-bringing-old-photos-back-to-life", "name": "Bringing Old Photos Back to Life", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open source", "licence": null, "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft model that restores old photos, repairing scratches and degradation with a separate stage that refines faces; code and weights under MIT.", "verified_on": "2026-10-15", "sources": [ "https://github.com/microsoft/Bringing-Old-Photos-Back-to-Life", "https://replicate.com/microsoft/bringing-old-photos-back-to-life" ] }, { "id": "falconsai-nsfw-image-detection", "name": "Falconsai NSFW image detection", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": null, "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vision Transformer classifier by Falcons.ai that flags not-safe-for-work images for content moderation; Apache 2.0.", "verified_on": "2026-10-15", "sources": [ "https://huggingface.co/Falconsai/nsfw_image_detection", "https://replicate.com/falcons-ai/nsfw_image_detection" ], "unverified": [] }, { "id": "megadescriptor", "name": "MegaDescriptor", "level": "model", "family_id": null, "developer": [ "czech-technical-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-11-15", "release_date_basis": "announcement page", "weights_date": "2023-11-06", "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Animal re-identification feature extractor introduced with the WildlifeDatasets toolkit, from the University of West Bohemia.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2311.09118", "https://huggingface.co/BVRA/MegaDescriptor-T-224", "https://huggingface.co/api/models?pipeline_tag=image-classification&sort=downloads&limit=50", "https://arxiv.org/html/2311.09118" ], "unverified": [] }, { "id": "rishit-dagli-plant-disease", "name": "Plant disease classifier (38 classes)", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Community plant leaf disease classifier for 38 crop and disease classes, published by Rishit Dagli on Kaggle.", "verified_on": "2026-10-12", "sources": [ "https://www.kaggle.com/models/rishitdagli/plant-disease" ], "unverified": [] }, { "id": "cppe-5-detectors", "name": "CPPE-5 medical PPE detectors", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Detectors (Faster R-CNN, SSD, YOLO, FCOS and others) trained on the CPPE-5 medical personal protective equipment dataset, published on Kaggle.", "verified_on": "2026-10-12", "sources": [ "https://www.kaggle.com/models/rishitdagli/faster-rcnn-cppe5" ], "unverified": [ "summary" ] }, { "id": "nts-net", "name": "NTS-Net", "level": "model", "family_id": null, "developer": [ "peking-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-09-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fine-grained bird classifier (Navigator-Teacher-Scrutinizer network) trained on CUB-200, in PyTorch Hub.", "verified_on": "2026-10-12", "sources": [ "https://github.com/nicolalandro/ntsnet-cub200", "https://pytorch.org/hub/nicolalandro_ntsnet-cub200_ntsnet", "https://raw.githubusercontent.com/pytorch/hub/master/nicolalandro_ntsnet-cub200_ntsnet.md", "https://arxiv.org/html/1809.00287", "https://arxiv.org/abs/1809.00287" ], "unverified": [] }, { "id": "sinet-portrait", "name": "SINet (portrait segmentation)", "level": "model", "family_id": null, "developer": [ "naver" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-12-16", "release_date_basis": "registry date", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NAVER Clova extremely lightweight portrait segmentation network.", "verified_on": "2026-10-12", "sources": [ "https://github.com/clovaai/ext_portrait_segmentation", "https://aihub.qualcomm.com/models/sinet" ], "unverified": [] }, { "id": "hbonet", "name": "HBONet", "level": "model", "family_id": null, "developer": [ "intel" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Harmonious bottleneck mobile classifier, in the OpenVINO Open Model Zoo.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/hbonet-0.25", "https://arxiv.org/html/1908.03888" ], "unverified": [] }, { "id": "6drepnet", "name": "6DRepNet", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-02-21", "release_date_basis": "registry date", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: Qualcomm AI Hub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Head pose estimation with a 6D rotation representation.", "verified_on": "2026-10-12", "sources": [ "https://github.com/thohemp/6DRepNet", "https://aihub.qualcomm.com/models/sixd_repnet" ], "unverified": [] }, { "id": "whenet", "name": "WHENet", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "Edge or toolkit builds listed in: NXP eIQ model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Wide-range head pose estimation network; NXP ships an i.MX recipe.", "verified_on": "2026-10-12", "sources": [ "https://github.com/NXP/eiq-model-zoo/tree/main/tasks/vision/pose-estimation" ], "unverified": [] }, { "id": "anti-spoof-mn3", "name": "Face anti-spoofing MobileNetV3", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Face anti-spoofing classifier on MobileNetV3, in the OpenVINO Open Model Zoo.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/anti-spoof-mn3" ], "unverified": [] }, { "id": "open-closed-eye", "name": "Open-closed-eye-0001", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small classifier for open or closed eyes, used for drowsiness detection; in the OpenVINO Open Model Zoo.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/open-closed-eye-0001" ], "unverified": [] }, { "id": "vehicle-reid-0001", "name": "Vehicle ReID 0001", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vehicle re-identification model in the OpenVINO Open Model Zoo.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/vehicle-reid-0001" ], "unverified": [] }, { "id": "background-matting-omz", "name": "Background matting (OpenVINO public models)", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "Edge or toolkit builds listed in: OpenVINO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Background and portrait matting models in the OpenVINO Open Model Zoo.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/background-matting-mobilenetv2" ], "unverified": [] }, { "id": "age-gender-cnn-onnx", "name": "Age and Gender Classification CNN", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "CNNs that classify age group and gender from face images, in the ONNX Model Zoo.", "verified_on": "2026-10-12", "sources": [ "https://github.com/onnx/models/tree/main/validated/vision/body_analysis/age_gender" ], "unverified": [] }, { "id": "roboflow-3-0", "name": "Roboflow 3.0", "level": "model", "family_id": null, "developer": [ "roboflow" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Roboflow 2.0 is the earlier generation.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Roboflow hosted training architecture for object detection, segmentation and classification used by many Roboflow Universe models.", "verified_on": "2026-10-12", "sources": [ "https://docs.roboflow.com/models/supported-models/roboflow-3.md" ], "unverified": [] }, { "id": "roboflow-instant", "name": "Roboflow Instant", "level": "model", "family_id": null, "developer": [ "roboflow" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Roboflow few-shot detection model that trains from a handful of labelled images.", "verified_on": "2026-10-12", "sources": [ "https://docs.roboflow.com/models/train/roboflow-instant.md" ], "unverified": [] }, { "id": "roboflow-yololite", "name": "YOLOlite", "level": "model", "family_id": null, "developer": [ "roboflow" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Lightweight YOLO detector supported for training and edge deployment on Roboflow.", "verified_on": "2026-10-12", "sources": [ "https://docs.roboflow.com/models/supported-models/yololite.md" ], "unverified": [] }, { "id": "edge-impulse-yolo-pro", "name": "YOLO-Pro", "level": "model", "family_id": null, "developer": [ "edge-impulse" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Pico to xlarge.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": "Edge or toolkit builds listed in: Edge Impulse.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Edge Impulse YOLO detector family sized from pico to xlarge for microcontrollers, NPUs and edge GPUs.", "verified_on": "2026-10-12", "sources": [ "https://docs.edgeimpulse.com/studio/projects/learning-blocks/blocks/object-detection/yolo-pro" ], "unverified": [] }, { "id": "gunadarma-university-oil-palm-detector", "name": "Oil palm detector (Universitas Gunadarma, Roboflow)", "level": "model", "family_id": null, "developer": [ "gunadarma-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Community oil palm detection project by Universitas Gunadarma on Roboflow Universe: about 2,380 images, class oilpalm, 4 trained models; architecture not stated on the search card.", "verified_on": "2026-10-12", "sources": [ "https://universe.roboflow.com/universitas-gunadarma/oil-palm-zcquy" ], "unverified": [ "summary", "architecture", "indonesian_measured", "indonesia_hosting" ] }, { "id": "gunadarma-oil-palm-health", "name": "Oil-Palm-Health (gunadarma, Roboflow)", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Community oil palm tree health detection project on Roboflow Universe: about 2,300 images with classes Dead, Grass, Healthy, Small and Yellow; 1 trained model.", "verified_on": "2026-10-12", "sources": [ "https://universe.roboflow.com/gunadarma/oil-palm-health-vglxy" ], "unverified": [ "summary", "architecture", "indonesian_measured", "indonesia_hosting" ] }, { "id": "uav-oil-palm", "name": "UAV oil palm (Roboflow)", "level": "model", "family_id": null, "developer": [], "model_class": "geospatial", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Community drone-image oil palm health detection project on Roboflow Universe: about 3,130 images with five health classes and 6 trained models.", "verified_on": "2026-10-12", "sources": [ "https://universe.roboflow.com/project-qryei/uav-oil-palm" ], "unverified": [ "summary", "architecture", "indonesian_measured", "indonesia_hosting" ] }, { "id": "binus-ptpn4-ganoderma-detection", "name": "Ganoderma Detection (Oil Palm), BINUS and PTPN IV", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Community Ganoderma basal stem rot detection project for oil palm on Roboflow Universe by a BINUS and PTPN IV workspace: about 475 images, 1 trained model.", "verified_on": "2026-10-12", "sources": [ "https://universe.roboflow.com/binus-ptpn4-ramadya/ganoderma-detection-oil-palm" ], "unverified": [ "summary", "architecture" ] }, { "id": "binus-oil-palm-ganoderma-detection", "name": "Oil Palm Ganoderma Detection (binus, Roboflow)", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Community oil palm Ganoderma detection project on Roboflow Universe: about 552 images, 2 trained models.", "verified_on": "2026-10-12", "sources": [ "https://universe.roboflow.com/binus-ozszb/oil-palm-ganoderma-detection" ], "unverified": [ "summary", "architecture" ] }, { "id": "zah-palm-oil-ripeness", "name": "Palm oil FFB ripeness (zah, Roboflow)", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Community oil palm fresh fruit bunch ripeness detection project on Roboflow Universe: about 7,070 images with Indonesian ripeness classes (mentah, kurang masak, masak, terlalu masak), 1 trained model.", "verified_on": "2026-10-12", "sources": [ "https://universe.roboflow.com/zah/palm-oil-fg85a" ], "unverified": [ "summary", "architecture", "indonesian_measured", "indonesia_hosting" ] }, { "id": "braincore-palm-oil-ripeness", "name": "Palm Oil Ripeness (braincore, Roboflow)", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Community oil palm fruit ripeness detection project on Roboflow Universe: about 3,940 images, 3 trained models.", "verified_on": "2026-10-12", "sources": [ "https://universe.roboflow.com/braincore/palm-oil-ripeness" ], "unverified": [ "summary", "architecture", "indonesian_measured", "indonesia_hosting" ] }, { "id": "sawit-ripeness-testing", "name": "Sawit ripeness (testing workspace, Roboflow)", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Community oil palm bunch grading project on Roboflow Universe: about 3,940 images with classes including Janjang Kosong, Kurang Masak, Masak, Mentah, Terlalu Masak and Abnormal; 2 trained models.", "verified_on": "2026-10-12", "sources": [ "https://universe.roboflow.com/testing-t85uq/sawit-svemx" ], "unverified": [ "summary", "architecture", "indonesian_measured", "indonesia_hosting" ] }, { "id": "google-bert", "name": "BERT", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Base (110M) and large (340M), cased and uncased, multilingual and Chinese checkpoints; many smaller official sizes.", "context_length": null, "release_date": "2018-10-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Checkpoint model cards list 103 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Burmese", "Javanese", "Malay", "Minangkabau", "Sundanese", "Tagalog", "Thai", "Vietnamese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Bidirectional transformer encoder pre-trained on unlabelled text with masked language modelling, then fine-tuned for classification, question answering and tagging.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1810.04805", "https://github.com/google-research/bert", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://huggingface.co/google-bert/bert-base-uncased", "https://huggingface.co/google-bert/bert-large-uncased" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "google-bert-base-uncased", "name": "BERT base uncased", "level": "variant", "family_id": "google-bert", "developer": [ "google" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 110106428, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BERT base uncased, a checkpoint of the BERT family. Bidirectional transformer encoder pre-trained on unlabelled text with masked language modelling, then fine-tuned for classification, question answering and tagging.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1810.04805", "https://huggingface.co/google-bert/bert-base-uncased", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "google-bert-large-uncased", "name": "BERT large uncased", "level": "variant", "family_id": "google-bert", "developer": [ "google" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 336226108, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BERT large uncased, a checkpoint of the BERT family. Bidirectional transformer encoder pre-trained on unlabelled text with masked language modelling, then fine-tuned for classification, question answering and tagging.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1810.04805", "https://huggingface.co/google-bert/bert-large-uncased", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "google-bert-base-multilingual-cased", "name": "BERT base multilingual cased", "level": "variant", "family_id": "google-bert", "developer": [ "google" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 178566653, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 104 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Burmese", "Javanese", "Malay", "Minangkabau", "Sundanese", "Tagalog", "Thai", "Vietnamese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Multilingual BERT base checkpoint pre-trained on Wikipedia in 104 languages, used as a starting point for non-English classification and tagging.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1810.04805", "https://huggingface.co/google-bert/bert-base-multilingual-cased", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "meta-roberta", "name": "RoBERTa", "level": "family", "family_id": null, "developer": [ "meta" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-07-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BERT-style encoder retrained with a more careful recipe (longer training, more data, no next-sentence objective) for text classification and extraction.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1907.11692", "https://github.com/facebookresearch/fairseq/tree/main/examples/roberta", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://pytorch.org/hub/pytorch_fairseq_roberta", "https://huggingface.co/FacebookAI/roberta-base", "https://huggingface.co/FacebookAI/roberta-large" ], "unverified": [] }, { "id": "meta-roberta-base", "name": "RoBERTa base", "level": "variant", "family_id": "meta-roberta", "developer": [ "meta" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 124697947, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "RoBERTa base, a checkpoint of the RoBERTa family. BERT-style encoder retrained with a more careful recipe (longer training, more data, no next-sentence objective) for text classification and extraction.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1907.11692", "https://huggingface.co/FacebookAI/roberta-base", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://pytorch.org/hub/pytorch_fairseq_roberta" ], "unverified": [] }, { "id": "meta-roberta-large", "name": "RoBERTa large", "level": "variant", "family_id": "meta-roberta", "developer": [ "meta" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 355412571, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "RoBERTa large, a checkpoint of the RoBERTa family. BERT-style encoder retrained with a more careful recipe (longer training, more data, no next-sentence objective) for text classification and extraction.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1907.11692", "https://huggingface.co/FacebookAI/roberta-large", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://pytorch.org/hub/pytorch_fairseq_roberta" ], "unverified": [] }, { "id": "meta-xlm-roberta", "name": "XLM-RoBERTa", "level": "family", "family_id": null, "developer": [ "meta" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-11-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "100 languages (paper); the model card language list includes Indonesian and Javanese.", "indonesian_support": "listed by developer", "regional_languages": [ "Javanese", "Khmer", "Lao", "Malay", "Burmese", "Sundanese", "Thai", "Tagalog", "Vietnamese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Multilingual RoBERTa encoder pre-trained on CommonCrawl text in 100 languages, used for cross-lingual classification, tagging and retrieval.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1911.02116", "https://modelzoo.co/model/xlm", "https://www.kaggle.com/models/keras/xlm_roberta", "https://huggingface.co/FacebookAI/xlm-roberta-base", "https://huggingface.co/FacebookAI/xlm-roberta-large" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "meta-xlm-roberta-base", "name": "XLM-RoBERTa base", "level": "variant", "family_id": "meta-xlm-roberta", "developer": [ "meta" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 278885778, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 94 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Javanese", "Khmer", "Lao", "Malay", "Burmese", "Sundanese", "Thai", "Tagalog", "Vietnamese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "XLM-RoBERTa base, a checkpoint of the XLM-RoBERTa family. Multilingual RoBERTa encoder pre-trained on CommonCrawl text in 100 languages, used for cross-lingual classification, tagging and retrieval.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1911.02116", "https://huggingface.co/FacebookAI/xlm-roberta-base", "https://modelzoo.co/model/xlm", "https://www.kaggle.com/models/keras/xlm_roberta" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "meta-xlm-roberta-large", "name": "XLM-RoBERTa large", "level": "variant", "family_id": "meta-xlm-roberta", "developer": [ "meta" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 561192082, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 94 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Javanese", "Khmer", "Lao", "Malay", "Burmese", "Sundanese", "Thai", "Tagalog", "Vietnamese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "XLM-RoBERTa large, a checkpoint of the XLM-RoBERTa family. Multilingual RoBERTa encoder pre-trained on CommonCrawl text in 100 languages, used for cross-lingual classification, tagging and retrieval.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1911.02116", "https://huggingface.co/FacebookAI/xlm-roberta-large", "https://modelzoo.co/model/xlm", "https://www.kaggle.com/models/keras/xlm_roberta" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "hugging-face-distilbert", "name": "DistilBERT", "level": "family", "family_id": null, "developer": [ "hugging-face" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-10-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Checkpoint model cards list 103 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Burmese", "Javanese", "Malay", "Minangkabau", "Sundanese", "Tagalog", "Thai", "Vietnamese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Smaller, faster BERT made by knowledge distillation during pre-training, keeping most of BERT's accuracy for on-device and low-cost text tasks.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1910.01108", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://aihub.qualcomm.com/models/distil_bert_base_uncased_hf", "https://huggingface.co/distilbert/distilbert-base-uncased", "https://huggingface.co/distilbert/distilbert-base-multilingual-cased" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "hugging-face-distilbert-base-uncased", "name": "DistilBERT base uncased", "level": "variant", "family_id": "hugging-face-distilbert", "developer": [ "hugging-face" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 66985530, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DistilBERT base uncased, a checkpoint of the DistilBERT family. Smaller, faster BERT made by knowledge distillation during pre-training, keeping most of BERT's accuracy for on-device and low-cost text tasks.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1910.01108", "https://huggingface.co/distilbert/distilbert-base-uncased", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://aihub.qualcomm.com/models/distil_bert_base_uncased_hf" ], "unverified": [] }, { "id": "hugging-face-distilbert-base-multilingual-cased", "name": "DistilBERT base multilingual cased", "level": "variant", "family_id": "hugging-face-distilbert", "developer": [ "hugging-face" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 135445755, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 104 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Burmese", "Javanese", "Malay", "Minangkabau", "Sundanese", "Tagalog", "Thai", "Vietnamese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "DistilBERT base multilingual cased, a checkpoint of the DistilBERT family. Smaller, faster BERT made by knowledge distillation during pre-training, keeping most of BERT's accuracy for on-device and low-cost text tasks.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1910.01108", "https://huggingface.co/distilbert/distilbert-base-multilingual-cased", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://aihub.qualcomm.com/models/distil_bert_base_uncased_hf" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "hugging-face-distilroberta-base", "name": "DistilRoBERTa base", "level": "variant", "family_id": "hugging-face-distilbert", "developer": [ "hugging-face" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 82760793, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DistilRoBERTa base, a checkpoint of the DistilBERT family. Smaller, faster BERT made by knowledge distillation during pre-training, keeping most of BERT's accuracy for on-device and low-cost text tasks.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1910.01108", "https://huggingface.co/distilbert/distilroberta-base", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://aihub.qualcomm.com/models/distil_bert_base_uncased_hf" ], "unverified": [] }, { "id": "google-albert", "name": "ALBERT", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-09-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Lite BERT that shares parameters across layers and factorises the embeddings to cut memory, for text classification and question answering.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1909.11942", "https://github.com/google-research/albert", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://aihub.qualcomm.com/models/albert_base_v2_hf", "https://huggingface.co/albert/albert-base-v2", "https://huggingface.co/albert/albert-xxlarge-v2" ], "unverified": [] }, { "id": "google-albert-base-v2", "name": "ALBERT base v2", "level": "variant", "family_id": "google-albert", "developer": [ "google" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 11842272, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ALBERT base v2, a checkpoint of the ALBERT family. Lite BERT that shares parameters across layers and factorises the embeddings to cut memory, for text classification and question answering.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1909.11942", "https://huggingface.co/albert/albert-base-v2", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://aihub.qualcomm.com/models/albert_base_v2_hf" ], "unverified": [] }, { "id": "google-albert-xxlarge-v2", "name": "ALBERT xxlarge v2", "level": "variant", "family_id": "google-albert", "developer": [ "google" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 223180256, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ALBERT xxlarge v2, a checkpoint of the ALBERT family. Lite BERT that shares parameters across layers and factorises the embeddings to cut memory, for text classification and question answering.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1909.11942", "https://huggingface.co/albert/albert-xxlarge-v2", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://aihub.qualcomm.com/models/albert_base_v2_hf" ], "unverified": [] }, { "id": "google-electra", "name": "ELECTRA", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-03-23", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Encoder pre-trained as a discriminator that detects replaced tokens, giving strong accuracy for its compute on classification and question answering.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2003.10555", "https://github.com/google-research/electra", "https://aihub.qualcomm.com/models/electra_bert_base_discrim_google", "https://huggingface.co/api/models?pipeline_tag=multiple-choice&sort=downloads&limit=50", "https://huggingface.co/google/electra-small-discriminator", "https://huggingface.co/google/electra-base-discriminator" ], "unverified": [] }, { "id": "google-electra-small-discriminator", "name": "ELECTRA small discriminator", "level": "variant", "family_id": "google-electra", "developer": [ "google" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ELECTRA small discriminator, a checkpoint of the ELECTRA family. Encoder pre-trained as a discriminator that detects replaced tokens, giving strong accuracy for its compute on classification and question answering.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2003.10555", "https://huggingface.co/google/electra-small-discriminator", "https://aihub.qualcomm.com/models/electra_bert_base_discrim_google", "https://huggingface.co/api/models?pipeline_tag=multiple-choice&sort=downloads&limit=50" ], "unverified": [] }, { "id": "google-electra-base-discriminator", "name": "ELECTRA base discriminator", "level": "variant", "family_id": "google-electra", "developer": [ "google" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ELECTRA base discriminator, a checkpoint of the ELECTRA family. Encoder pre-trained as a discriminator that detects replaced tokens, giving strong accuracy for its compute on classification and question answering.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2003.10555", "https://huggingface.co/google/electra-base-discriminator", "https://aihub.qualcomm.com/models/electra_bert_base_discrim_google", "https://huggingface.co/api/models?pipeline_tag=multiple-choice&sort=downloads&limit=50" ], "unverified": [] }, { "id": "google-electra-large-discriminator", "name": "ELECTRA large discriminator", "level": "variant", "family_id": "google-electra", "developer": [ "google" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ELECTRA large discriminator, a checkpoint of the ELECTRA family. Encoder pre-trained as a discriminator that detects replaced tokens, giving strong accuracy for its compute on classification and question answering.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2003.10555", "https://huggingface.co/google/electra-large-discriminator", "https://aihub.qualcomm.com/models/electra_bert_base_discrim_google", "https://huggingface.co/api/models?pipeline_tag=multiple-choice&sort=downloads&limit=50" ], "unverified": [] }, { "id": "microsoft-deberta", "name": "DeBERTa", "level": "family", "family_id": null, "developer": [ "microsoft" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "v1, v2 and v3 lines; the multilingual mDeBERTa-v3 base is a separate AURION record (microsoft-mdeberta-v3-base).", "context_length": null, "release_date": "2020-06-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Encoder with disentangled attention over content and position; version 3 uses ELECTRA-style pre-training and is a common base for classification and NLI models.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2006.03654", "https://github.com/microsoft/DeBERTa", "https://www.kaggle.com/models/keras/deberta_v3", "https://paperswithcode.com/method/deberta", "https://huggingface.co/microsoft/deberta-v3-base", "https://huggingface.co/microsoft/deberta-v3-large" ], "unverified": [] }, { "id": "microsoft-deberta-v3-base", "name": "DeBERTa-v3 base", "level": "variant", "family_id": "microsoft-deberta", "developer": [ "microsoft" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-11-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DeBERTa-v3 base, a checkpoint of the DeBERTa family. Encoder with disentangled attention over content and position; version 3 uses ELECTRA-style pre-training and is a common base for classification and NLI models.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2111.09543", "https://huggingface.co/microsoft/deberta-v3-base", "https://www.kaggle.com/models/keras/deberta_v3", "https://paperswithcode.com/method/deberta" ], "unverified": [] }, { "id": "microsoft-deberta-v3-large", "name": "DeBERTa-v3 large", "level": "variant", "family_id": "microsoft-deberta", "developer": [ "microsoft" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-11-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DeBERTa-v3 large, a checkpoint of the DeBERTa family. Encoder with disentangled attention over content and position; version 3 uses ELECTRA-style pre-training and is a common base for classification and NLI models.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2111.09543", "https://huggingface.co/microsoft/deberta-v3-large", "https://www.kaggle.com/models/keras/deberta_v3", "https://paperswithcode.com/method/deberta" ], "unverified": [] }, { "id": "google-mobilebert", "name": "MobileBERT", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-04-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU", "CPU server" ], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Compact, task-agnostic BERT built for phones, distilled from a large teacher; used for on-device classification and question answering.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2004.02984", "https://github.com/google-research/google-research/tree/master/mobilebert", "https://huggingface.co/google/mobilebert-uncased", "https://aihub.qualcomm.com/models/mobile_bert_uncased_google", "https://huggingface.co/api/models?pipeline_tag=question-answering&sort=downloads&limit=50" ], "unverified": [] }, { "id": "allen-ai-longformer", "name": "Longformer", "level": "family", "family_id": null, "developer": [ "allen-ai" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-04-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Transformer with sliding-window plus global attention that scales linearly with length, for documents of thousands of tokens; LED is its encoder-decoder for long summarisation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2004.05150", "https://github.com/allenai/longformer", "https://paperswithcode.com/method/longformer", "https://github.com/tensorflow/models/tree/master/official/projects/longformer", "https://huggingface.co/allenai/longformer-base-4096", "https://huggingface.co/allenai/led-base-16384" ], "unverified": [] }, { "id": "allen-ai-longformer-base-4096", "name": "Longformer base 4096", "level": "variant", "family_id": "allen-ai-longformer", "developer": [ "allen-ai" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Longformer base 4096, a checkpoint of the Longformer family. Transformer with sliding-window plus global attention that scales linearly with length, for documents of thousands of tokens; LED is its encoder-decoder for long summarisation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2004.05150", "https://huggingface.co/allenai/longformer-base-4096", "https://paperswithcode.com/method/longformer", "https://github.com/tensorflow/models/tree/master/official/projects/longformer" ], "unverified": [] }, { "id": "allen-ai-led-base-16384", "name": "LED base 16384", "level": "variant", "family_id": "allen-ai-longformer", "developer": [ "allen-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Longformer Encoder-Decoder base checkpoint for generating summaries of inputs up to 16,384 tokens.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2004.05150", "https://huggingface.co/allenai/led-base-16384", "https://paperswithcode.com/method/longformer", "https://github.com/tensorflow/models/tree/master/official/projects/longformer" ], "unverified": [] }, { "id": "microsoft-minilm", "name": "MiniLM", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-02-25", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small encoder compressed from a large transformer by deep self-attention distillation; the base of many sentence-embedding and reranking models.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2002.10957", "https://github.com/microsoft/unilm/tree/master/minilm", "https://huggingface.co/microsoft/MiniLM-L12-H384-uncased", "https://huggingface.co/api/models?pipeline_tag=sentence-similarity&sort=downloads&limit=50" ], "unverified": [] }, { "id": "microsoft-mpnet-base", "name": "MPNet base", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 132993790, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2020-04-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Encoder pre-trained with masked and permuted language modelling; the base of the widely used all-mpnet-base-v2 sentence embedder.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2004.09297", "https://huggingface.co/microsoft/mpnet-base", "https://github.com/onnx/models/tree/main/Natural_Language_Processing/mpnet_Opset16_transformers" ], "unverified": [] }, { "id": "google-cmu-xlnet", "name": "XLNet", "level": "model", "family_id": null, "developer": [ "google", "carnegie-mellon-university" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-06-19", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Generalised autoregressive pre-training over permutations of word order, built on Transformer-XL, for classification and question answering.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1906.08237", "https://github.com/zihangdai/xlnet", "https://huggingface.co/xlnet/xlnet-base-cased", "https://pytorch.org/hub/huggingface_pytorch-transformers" ], "unverified": [] }, { "id": "google-cmu-transformer-xl", "name": "Transformer-XL", "level": "model", "family_id": null, "developer": [ "google", "carnegie-mellon-university" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-01-09", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Language model with segment-level recurrence and relative positions, which lets it use context beyond a fixed length.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1901.02860", "https://github.com/kimiyoung/transformer-xl", "https://pytorch.org/hub/huggingface_pytorch-transformers", "https://modelzoo.co/model/transformer-xl" ], "unverified": [] }, { "id": "google-bigbird", "name": "BigBird", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-07-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Transformer with sparse (random, window and global) attention that handles sequences up to 4,096 tokens, for long-document question answering and summarisation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2007.14062", "https://github.com/google-research/bigbird", "https://huggingface.co/google/bigbird-roberta-base", "https://github.com/tensorflow/models/tree/master/official/projects/bigbird", "https://github.com/onnx/models/tree/main/Natural_Language_Processing/bigbird_Opset16_transformers" ], "unverified": [] }, { "id": "inria-camembert", "name": "CamemBERT", "level": "model", "family_id": null, "developer": [ "inria" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 111246085, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2019-11-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "French", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "French RoBERTa-style encoder trained on French web text, for French tagging, parsing and classification.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1911.03894", "https://huggingface.co/almanach/camembert-base", "https://github.com/onnx/models/tree/main/Natural_Language_Processing/camembert_Opset16_transformers", "https://huggingface.co/api/models?pipeline_tag=fill-mask&sort=downloads&limit=50" ], "unverified": [] }, { "id": "jhu-clsp-mmbert", "name": "mmBERT", "level": "family", "family_id": null, "developer": [ "johns-hopkins-university" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-07-23", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Checkpoint model cards list 1811 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Acehnese", "Balinese", "Banjar", "Buginese", "Javanese", "Madurese", "Minangkabau", "Sundanese", "Malay" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Modern multilingual encoder trained on text in over 1,800 languages, a ModernBERT-style successor to XLM-R for classification and retrieval.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/api/models?pipeline_tag=fill-mask&sort=likes&limit=50", "https://huggingface.co/jhu-clsp/mmBERT-small", "https://huggingface.co/jhu-clsp/mmBERT-base" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "jhu-clsp-mmbert-small", "name": "mmBERT small", "level": "variant", "family_id": "jhu-clsp-mmbert", "developer": [ "johns-hopkins-university" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-08-21", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 1811 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Acehnese", "Balinese", "Banjar", "Buginese", "Javanese", "Madurese", "Minangkabau", "Sundanese", "Malay" ], "indonesia_hosting": "unknown", "status": "active", "summary": "mmBERT small, a checkpoint of the mmBERT family. Modern multilingual encoder trained on text in over 1,800 languages, a ModernBERT-style successor to XLM-R for classification and retrieval.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/jhu-clsp/mmBERT-small", "https://huggingface.co/api/models?pipeline_tag=fill-mask&sort=likes&limit=50" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "jhu-clsp-mmbert-base", "name": "mmBERT base", "level": "variant", "family_id": "jhu-clsp-mmbert", "developer": [ "johns-hopkins-university" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-07-23", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 1811 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Acehnese", "Balinese", "Banjar", "Buginese", "Javanese", "Madurese", "Minangkabau", "Sundanese", "Malay" ], "indonesia_hosting": "unknown", "status": "active", "summary": "mmBERT base, a checkpoint of the mmBERT family. Modern multilingual encoder trained on text in over 1,800 languages, a ModernBERT-style successor to XLM-R for classification and retrieval.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/jhu-clsp/mmBERT-base", "https://huggingface.co/api/models?pipeline_tag=fill-mask&sort=likes&limit=50" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "jhu-clsp-ettin", "name": "Ettin", "level": "family", "family_id": null, "developer": [ "johns-hopkins-university" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03-24", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Paired suite of encoder-only and decoder-only models from 17M to 1B parameters trained on the same data, to compare the two designs.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/api/models?pipeline_tag=visual-document-retrieval&sort=likes&limit=50", "https://huggingface.co/jhu-clsp/ettin-encoder-150m", "https://huggingface.co/jhu-clsp/ettin-encoder-1b" ], "unverified": [] }, { "id": "jhu-clsp-ettin-encoder-150m", "name": "Ettin encoder 150M", "level": "variant", "family_id": "jhu-clsp-ettin", "developer": [ "johns-hopkins-university" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03-24", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Ettin encoder 150M, a checkpoint of the Ettin family. Paired suite of encoder-only and decoder-only models from 17M to 1B parameters trained on the same data, to compare the two designs.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/jhu-clsp/ettin-encoder-150m", "https://huggingface.co/api/models?pipeline_tag=visual-document-retrieval&sort=likes&limit=50" ], "unverified": [] }, { "id": "jhu-clsp-ettin-encoder-1b", "name": "Ettin encoder 1B", "level": "variant", "family_id": "jhu-clsp-ettin", "developer": [ "johns-hopkins-university" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-07-03", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Ettin encoder 1B, a checkpoint of the Ettin family. Paired suite of encoder-only and decoder-only models from 17M to 1B parameters trained on the same data, to compare the two designs.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/jhu-clsp/ettin-encoder-1b", "https://huggingface.co/api/models?pipeline_tag=visual-document-retrieval&sort=likes&limit=50" ], "unverified": [] }, { "id": "google-muril", "name": "MuRIL", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-03-19", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "17 Indian languages and English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BERT encoder pre-trained on 17 Indian languages and their transliterations, for Indian-language classification and tagging.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.10730", "https://huggingface.co/google/muril-base-cased", "https://www.kaggle.com/models/google/muril" ], "unverified": [] }, { "id": "allen-ai-elmo", "name": "ELMo", "level": "model", "family_id": null, "developer": [ "allen-ai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "Bidirectional LSTM language model", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-02-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Deep contextual word representations from a bidirectional LSTM language model, used as input features for NLP tasks.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1802.05365", "https://allenai.org/allennlp/software/elmo", "https://www.kaggle.com/models/google/elmo", "https://paperswithcode.com/method/elmo" ], "unverified": [] }, { "id": "meta-fasttext", "name": "fastText word vectors", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "Shallow skip-gram and CBOW with character n-grams", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2016-07-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-sa-3-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "157 languages, including Indonesian (cc.id vectors)", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Word vectors that add character n-grams to the skip-gram model, with pre-trained vectors for 157 languages, used for lightweight text classification and similarity.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1607.04606", "https://fasttext.cc/docs/en/crawl-vectors.html", "https://huggingface.co/facebook/fasttext-id-vectors", "https://modelzoo.co/model/tensorflow-fasttext", "https://paperswithcode.com/method/glove" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "stanford-university-glove", "name": "GloVe", "level": "model", "family_id": null, "developer": [ "stanford-university" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "Global log-bilinear regression on word co-occurrence counts", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2014-10", "release_date_basis": "inferred from page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Pre-trained English word vectors learned from global word co-occurrence statistics, used as features for text models.", "verified_on": "2026-10-12", "sources": [ "https://nlp.stanford.edu/projects/glove/", "https://aclanthology.org/D14-1162", "https://modelzoo.co/model/tensorflow-fasttext", "https://paperswithcode.com/method/glove" ], "unverified": [] }, { "id": "meta-laser", "name": "LASER", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "BiLSTM encoder shared across languages", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-12-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Language-agnostic sentence embeddings from one encoder covering 93 languages, for bitext mining and cross-lingual classification.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1812.10464", "https://github.com/facebookresearch/LASER", "https://modelzoo.co/model/laser" ], "unverified": [] }, { "id": "microsoft-biomedbert", "name": "BiomedBERT (PubMedBERT)", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-07-31", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BERT pre-trained from scratch on PubMed abstracts and full-text articles, for biomedical entity recognition, relation extraction and question answering.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2007.15779", "https://huggingface.co/microsoft/BiomedNLP-BiomedBERT-base-uncased-abstract-fulltext", "https://ai.azure.com/catalog/models?industry=health-and-life-sciences", "https://huggingface.co/api/models?pipeline_tag=fill-mask&sort=downloads&limit=50" ], "unverified": [] }, { "id": "emilyalsentzer-bio-clinicalbert", "name": "Bio_ClinicalBERT", "level": "model", "family_id": null, "developer": [ "emilyalsentzer" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-04-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BioBERT further trained on MIMIC-III clinical notes, for clinical text classification and entity recognition.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1904.03323", "https://huggingface.co/emilyalsentzer/Bio_ClinicalBERT", "https://huggingface.co/api/models?pipeline_tag=fill-mask&sort=downloads&limit=50" ], "unverified": [] }, { "id": "korea-university-biobert", "name": "BioBERT", "level": "model", "family_id": null, "developer": [ "korea-university" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-01-25", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BERT further pre-trained on PubMed and PMC text for biomedical named entity recognition, relation extraction and question answering.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1901.08746", "https://github.com/dmis-lab/biobert", "https://huggingface.co/dmis-lab/biobert-v1.1", "https://catalog.ngc.nvidia.com/orgs/nvidia/models/biobertbasecasedfornemo" ], "unverified": [] }, { "id": "cambridge-ltl-sapbert", "name": "SapBERT", "level": "model", "family_id": null, "developer": [ "university-of-cambridge" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 109482752, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2020-10-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "PubMedBERT trained with self-alignment on UMLS synonyms so that names of the same medical concept get close embeddings, for medical entity linking.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2010.11784", "https://huggingface.co/cambridgeltl/SapBERT-from-PubMedBERT-fulltext", "https://huggingface.co/api/models?pipeline_tag=feature-extraction&sort=downloads&limit=50" ], "unverified": [] }, { "id": "aueb-legal-bert", "name": "LEGAL-BERT", "level": "model", "family_id": null, "developer": [ "athens-university-of-economics-and-business" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-10-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-sa-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BERT pre-trained on English legislation, court cases and contracts, for legal text classification and extraction.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2010.02559", "https://huggingface.co/nlpaueb/legal-bert-base-uncased", "https://huggingface.co/api/models?pipeline_tag=fill-mask&sort=downloads&limit=50" ], "unverified": [] }, { "id": "prosus-ai-finbert", "name": "FinBERT (ProsusAI)", "level": "model", "family_id": null, "developer": [ "prosus-ai" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-08-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BERT further trained on financial text and fine-tuned to classify the sentiment of financial news as positive, negative or neutral.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1908.10063", "https://huggingface.co/ProsusAI/finbert", "https://huggingface.co/api/models?pipeline_tag=text-classification&sort=downloads&limit=50" ], "unverified": [] }, { "id": "stanford-university-deidentifier-base", "name": "Stanford De-identifier base", "level": "model", "family_id": null, "developer": [ "stanford-university" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-06-16", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Token classifier that finds protected health information in radiology and clinical notes so it can be removed before data sharing.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/StanfordAIMI/stanford-deidentifier-base", "https://huggingface.co/api/models?pipeline_tag=token-classification&sort=downloads&limit=50" ], "unverified": [] }, { "id": "dslim-bert-base-ner", "name": "bert-base-NER", "level": "model", "family_id": null, "developer": [ "dslim" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 108317193, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BERT base fine-tuned on CoNLL-2003 to tag person, organisation, location and miscellaneous names in English text.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/dslim/bert-base-NER", "https://huggingface.co/api/models?pipeline_tag=token-classification&sort=downloads&limit=50", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/bert-base-ner" ], "unverified": [] }, { "id": "cardiff-nlp-twitter-roberta-base-sentiment-latest", "name": "Twitter-RoBERTa base sentiment (latest)", "level": "model", "family_id": null, "developer": [ "cardiff-nlp" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-03-15", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "RoBERTa base trained on about 124 million tweets and fine-tuned for English tweet sentiment, used for social media monitoring.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/cardiffnlp/twitter-roberta-base-sentiment-latest", "https://huggingface.co/api/models?pipeline_tag=text-classification&sort=downloads&limit=50" ], "unverified": [] }, { "id": "microsoft-llmlingua-2", "name": "LLMLingua-2", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 558945282, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-03-17", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Token classifier, distilled from GPT-4 labels, that drops low-value tokens to compress prompts before they are sent to a large model.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/microsoft/llmlingua-2-xlm-roberta-large-meetingbank", "https://huggingface.co/api/models?pipeline_tag=token-classification&sort=downloads&limit=50", "https://www.modelscope.cn/models/microsoft/llmlingua-2-xlm-roberta-large-meetingbank" ], "unverified": [] }, { "id": "urchade-gliner", "name": "GLiNER", "level": "family", "family_id": null, "developer": [ "urchade" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-11-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: multilingual, en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Bidirectional encoder that extracts entities of any type named at run time, a small alternative to large models for open named entity recognition.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2311.08526", "https://github.com/urchade/GLiNER", "https://huggingface.co/api/models?pipeline_tag=token-classification&sort=likes&limit=50", "https://www.modelscope.cn/models/AI-ModelScope/gliner_multi", "https://huggingface.co/urchade/gliner_multi-v2.1", "https://huggingface.co/urchade/gliner_medium-v2.1" ], "unverified": [] }, { "id": "urchade-gliner-multi-v2-1", "name": "GLiNER multi v2.1", "level": "variant", "family_id": "urchade-gliner", "developer": [ "urchade" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 288949504, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-04-09", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "GLiNER multi v2.1, a checkpoint of the GLiNER family. Bidirectional encoder that extracts entities of any type named at run time, a small alternative to large models for open named entity recognition.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2311.08526", "https://huggingface.co/urchade/gliner_multi-v2.1", "https://huggingface.co/api/models?pipeline_tag=token-classification&sort=likes&limit=50", "https://www.modelscope.cn/models/AI-ModelScope/gliner_multi" ], "unverified": [] }, { "id": "urchade-gliner-medium-v2-1", "name": "GLiNER medium v2.1", "level": "variant", "family_id": "urchade-gliner", "developer": [ "urchade" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 195175936, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-03-17", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "GLiNER medium v2.1, a checkpoint of the GLiNER family. Bidirectional encoder that extracts entities of any type named at run time, a small alternative to large models for open named entity recognition.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2311.08526", "https://huggingface.co/urchade/gliner_medium-v2.1", "https://huggingface.co/api/models?pipeline_tag=token-classification&sort=likes&limit=50", "https://www.modelscope.cn/models/AI-ModelScope/gliner_multi" ], "unverified": [] }, { "id": "flax-community-indonesian-roberta-base", "name": "Indonesian RoBERTa base", "level": "model", "family_id": null, "developer": [ "flax-community" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 124697947, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Indonesian", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "RoBERTa base masked language model trained on Indonesian text, a starting point for Indonesian classification and tagging.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/flax-community/indonesian-roberta-base", "https://huggingface.co/api/models?pipeline_tag=token-classification&sort=downloads&limit=50" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "baai-bge-v1-5", "name": "BGE v1.5", "level": "family", "family_id": null, "developer": [ "baai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Small, base and large sizes in English and Chinese; BGE-M3 and other BGE records already in AURION.", "context_length": null, "release_date": "2023-09-11", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, zh", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "English and Chinese text embedding models from BAAI's C-Pack work, for retrieval and semantic search; v1.5 improves the similarity distribution.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2309.07597", "https://github.com/FlagOpen/FlagEmbedding", "https://huggingface.co/api/models?pipeline_tag=feature-extraction&sort=downloads&limit=50", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://huggingface.co/BAAI/bge-small-en-v1.5", "https://huggingface.co/BAAI/bge-base-en-v1.5" ], "unverified": [] }, { "id": "baai-bge-small-en-v1-5", "name": "bge-small-en-v1.5", "level": "variant", "family_id": "baai-bge-v1-5", "developer": [ "baai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 33360512, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-09-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "bge-small-en-v1.5, a checkpoint of the BGE v1.5 family. English and Chinese text embedding models from BAAI's C-Pack work, for retrieval and semantic search; v1.5 improves the similarity distribution.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2309.07597", "https://huggingface.co/BAAI/bge-small-en-v1.5", "https://huggingface.co/api/models?pipeline_tag=feature-extraction&sort=downloads&limit=50", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "baai-bge-base-en-v1-5", "name": "bge-base-en-v1.5", "level": "variant", "family_id": "baai-bge-v1-5", "developer": [ "baai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 109482752, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-09-11", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "bge-base-en-v1.5, a checkpoint of the BGE v1.5 family. English and Chinese text embedding models from BAAI's C-Pack work, for retrieval and semantic search; v1.5 improves the similarity distribution.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2309.07597", "https://huggingface.co/BAAI/bge-base-en-v1.5", "https://huggingface.co/api/models?pipeline_tag=feature-extraction&sort=downloads&limit=50", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "baai-bge-large-en-v1-5", "name": "bge-large-en-v1.5", "level": "variant", "family_id": "baai-bge-v1-5", "developer": [ "baai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 335142400, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-09-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "bge-large-en-v1.5, a checkpoint of the BGE v1.5 family. English and Chinese text embedding models from BAAI's C-Pack work, for retrieval and semantic search; v1.5 improves the similarity distribution.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2309.07597", "https://huggingface.co/BAAI/bge-large-en-v1.5", "https://huggingface.co/api/models?pipeline_tag=feature-extraction&sort=downloads&limit=50", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "baai-bge-large-zh-v1-5", "name": "bge-large-zh-v1.5", "level": "variant", "family_id": "baai-bge-v1-5", "developer": [ "baai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-09-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Chinese", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "bge-large-zh-v1.5, a checkpoint of the BGE v1.5 family. English and Chinese text embedding models from BAAI's C-Pack work, for retrieval and semantic search; v1.5 improves the similarity distribution.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2309.07597", "https://huggingface.co/BAAI/bge-large-zh-v1.5", "https://huggingface.co/api/models?pipeline_tag=feature-extraction&sort=downloads&limit=50", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "baai-bge-reranker", "name": "BGE Reranker (v1)", "level": "family", "family_id": null, "developer": [ "baai" ], "model_class": "reranker", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Base and large; the later bge-reranker-v2-m3 is a separate AURION record.", "context_length": null, "release_date": "2023-09-11", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, zh", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Cross-encoders that score a query and passage together to re-rank retrieval results, in English and Chinese.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2309.07597", "https://huggingface.co/api/models?pipeline_tag=feature-extraction&sort=downloads&limit=50", "https://www.modelscope.cn/models/BAAI/bge-reranker-large", "https://huggingface.co/BAAI/bge-reranker-base", "https://huggingface.co/BAAI/bge-reranker-large" ], "unverified": [] }, { "id": "baai-bge-reranker-base", "name": "bge-reranker-base", "level": "variant", "family_id": "baai-bge-reranker", "developer": [ "baai" ], "model_class": "reranker", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 278044931, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-09-11", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, zh", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "bge-reranker-base, a checkpoint of the BGE Reranker (v1) family. Cross-encoders that score a query and passage together to re-rank retrieval results, in English and Chinese.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2309.07597", "https://huggingface.co/BAAI/bge-reranker-base", "https://huggingface.co/api/models?pipeline_tag=feature-extraction&sort=downloads&limit=50", "https://www.modelscope.cn/models/BAAI/bge-reranker-large" ], "unverified": [] }, { "id": "baai-bge-reranker-large", "name": "bge-reranker-large", "level": "variant", "family_id": "baai-bge-reranker", "developer": [ "baai" ], "model_class": "reranker", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 559891971, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-09-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, zh", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "bge-reranker-large, a checkpoint of the BGE Reranker (v1) family. Cross-encoders that score a query and passage together to re-rank retrieval results, in English and Chinese.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2309.07597", "https://huggingface.co/BAAI/bge-reranker-large", "https://huggingface.co/api/models?pipeline_tag=feature-extraction&sort=downloads&limit=50", "https://www.modelscope.cn/models/BAAI/bge-reranker-large" ], "unverified": [] }, { "id": "alibaba-gte", "name": "GTE (General Text Embeddings)", "level": "family", "family_id": null, "developer": [ "alibaba" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "thenlper gte-small/base/large (2023), gte-v1.5 English (2024), gte-Qwen2 1.5B and 7B instruct LLM-based embedders; gte-multilingual-base is a separate AURION record.", "context_length": null, "release_date": "2023-07-27", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba text embedding line trained with multi-stage contrastive learning, for retrieval, clustering and semantic similarity.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2308.03281", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://huggingface.co/api/models?pipeline_tag=sentence-similarity&sort=downloads&limit=50", "https://huggingface.co/thenlper/gte-large", "https://huggingface.co/thenlper/gte-small" ], "unverified": [] }, { "id": "alibaba-gte-large", "name": "gte-large", "level": "variant", "family_id": "alibaba-gte", "developer": [ "alibaba" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 335142400, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-07-27", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "gte-large, a checkpoint of the GTE (General Text Embeddings) family. Alibaba text embedding line trained with multi-stage contrastive learning, for retrieval, clustering and semantic similarity.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2308.03281", "https://huggingface.co/thenlper/gte-large", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://huggingface.co/api/models?pipeline_tag=sentence-similarity&sort=downloads&limit=50" ], "unverified": [] }, { "id": "alibaba-gte-small", "name": "gte-small", "level": "variant", "family_id": "alibaba-gte", "developer": [ "alibaba" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 33360512, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-07-27", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "gte-small, a checkpoint of the GTE (General Text Embeddings) family. Alibaba text embedding line trained with multi-stage contrastive learning, for retrieval, clustering and semantic similarity.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2308.03281", "https://huggingface.co/thenlper/gte-small", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://huggingface.co/api/models?pipeline_tag=sentence-similarity&sort=downloads&limit=50" ], "unverified": [] }, { "id": "alibaba-gte-large-en-v1-5", "name": "gte-large-en-v1.5", "level": "variant", "family_id": "alibaba-gte", "developer": [ "alibaba" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 434139136, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-04-20", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "gte-large-en-v1.5, a checkpoint of the GTE (General Text Embeddings) family. Alibaba text embedding line trained with multi-stage contrastive learning, for retrieval, clustering and semantic similarity.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2308.03281", "https://huggingface.co/Alibaba-NLP/gte-large-en-v1.5", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://huggingface.co/api/models?pipeline_tag=sentence-similarity&sort=downloads&limit=50" ], "unverified": [] }, { "id": "alibaba-gte-qwen2-7b-instruct", "name": "gte-Qwen2-7B-instruct", "level": "variant", "family_id": "alibaba-gte", "developer": [ "alibaba" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 7612620288, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-06-15", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "gte-Qwen2-7B-instruct, a checkpoint of the GTE (General Text Embeddings) family. Alibaba text embedding line trained with multi-stage contrastive learning, for retrieval, clustering and semantic similarity.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2308.03281", "https://huggingface.co/Alibaba-NLP/gte-Qwen2-7B-instruct", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://huggingface.co/api/models?pipeline_tag=sentence-similarity&sort=downloads&limit=50" ], "unverified": [] }, { "id": "alibaba-gte-multilingual-reranker-base", "name": "gte-multilingual-reranker-base", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "reranker", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 305959681, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-07-20", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 75 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Javanese", "Khmer", "Lao", "Malay", "Burmese", "Thai", "Tagalog", "Vietnamese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Multilingual cross-encoder reranker from the GTE line for re-ordering retrieved passages in over 70 languages.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Alibaba-NLP/gte-multilingual-reranker-base", "https://huggingface.co/api/models?pipeline_tag=text-ranking&sort=downloads&limit=50" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "alibaba-gte-reranker-modernbert-base", "name": "gte-reranker-modernbert-base", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "reranker", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 149605633, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-01-20", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "English cross-encoder reranker built on ModernBERT base, for re-ranking search results.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Alibaba-NLP/gte-reranker-modernbert-base", "https://www.modelscope.cn/models/iic/gte-reranker-modernbert-base" ], "unverified": [] }, { "id": "microsoft-multilingual-e5-small", "name": "multilingual-e5-small", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 117654272, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-06-30", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 94 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Javanese", "Khmer", "Lao", "Malay", "Burmese", "Sundanese", "Thai", "Tagalog", "Vietnamese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Small multilingual E5 text embedding model for retrieval and similarity across about 100 languages.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/intfloat/multilingual-e5-small", "https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/maas/use-open-models" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "nomic-ai-nomic-embed-text-v1-family", "name": "Nomic Embed Text v1", "level": "family", "family_id": null, "developer": [ "nomic-ai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-01-31", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fully open long-context English text embedder (8,192 tokens) with released training data and code; v1.5 adds Matryoshka resizable vectors.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2402.01613", "https://huggingface.co/api/models?pipeline_tag=sentence-similarity&sort=downloads&limit=50", "https://aihub.qualcomm.com/models/nomic_embed_text", "https://huggingface.co/nomic-ai/nomic-embed-text-v1", "https://huggingface.co/nomic-ai/nomic-embed-text-v1.5" ], "unverified": [] }, { "id": "nomic-ai-nomic-embed-text-v1", "name": "nomic-embed-text-v1", "level": "variant", "family_id": "nomic-ai-nomic-embed-text-v1-family", "developer": [ "nomic-ai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 136731648, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-01-31", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "nomic-embed-text-v1, a checkpoint of the Nomic Embed Text v1 family. Fully open long-context English text embedder (8,192 tokens) with released training data and code; v1.5 adds Matryoshka resizable vectors.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2402.01613", "https://huggingface.co/nomic-ai/nomic-embed-text-v1", "https://huggingface.co/api/models?pipeline_tag=sentence-similarity&sort=downloads&limit=50", "https://aihub.qualcomm.com/models/nomic_embed_text" ], "unverified": [] }, { "id": "nomic-ai-nomic-embed-text-v1-5", "name": "nomic-embed-text-v1.5", "level": "variant", "family_id": "nomic-ai-nomic-embed-text-v1-family", "developer": [ "nomic-ai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 136731648, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-02-10", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "nomic-embed-text-v1.5, a checkpoint of the Nomic Embed Text v1 family. Fully open long-context English text embedder (8,192 tokens) with released training data and code; v1.5 adds Matryoshka resizable vectors.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2402.01613", "https://huggingface.co/nomic-ai/nomic-embed-text-v1.5", "https://huggingface.co/api/models?pipeline_tag=sentence-similarity&sort=downloads&limit=50", "https://aihub.qualcomm.com/models/nomic_embed_text" ], "unverified": [] }, { "id": "nomic-ai-nomic-embed-vision-v1-5", "name": "nomic-embed-vision-v1.5", "level": "model", "family_id": null, "developer": [ "nomic-ai" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 92946688, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-06-01", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Image encoder aligned to the nomic-embed-text-v1.5 space, so images and text can be searched in one vector index.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/nomic-ai/nomic-embed-vision-v1.5", "https://huggingface.co/api/models?pipeline_tag=image-feature-extraction&sort=downloads&limit=50", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "jina-ai-jina-embeddings-v2-base-en", "name": "jina-embeddings-v2-base-en", "level": "model", "family_id": null, "developer": [ "jina-ai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 137368320, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-09-27", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "English text embedder that accepts 8,192-token inputs using ALiBi attention, for long-document retrieval.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/jinaai/jina-embeddings-v2-base-en", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://huggingface.co/api/models?pipeline_tag=visual-document-retrieval&sort=downloads&limit=50" ], "unverified": [] }, { "id": "jina-ai-jina-embeddings-v4", "name": "jina-embeddings-v4", "level": "model", "family_id": null, "developer": [ "jina-ai" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 3754885248, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-05-07", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": "qwen-research-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multimodal, multilingual embedder built on Qwen2.5-VL for text and visually rich documents, with single-vector and multi-vector outputs.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/jinaai/jina-embeddings-v4", "https://huggingface.co/api/models?pipeline_tag=visual-document-retrieval&sort=downloads&limit=50" ], "unverified": [] }, { "id": "jina-ai-jina-reranker-v2-base-multilingual", "name": "jina-reranker-v2-base-multilingual", "level": "model", "family_id": null, "developer": [ "jina-ai" ], "model_class": "reranker", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 278437633, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-06-19", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multilingual cross-encoder reranker that also handles function-calling and code search queries.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/jinaai/jina-reranker-v2-base-multilingual", "https://huggingface.co/api/models?pipeline_tag=text-ranking&sort=downloads&limit=50" ], "unverified": [] }, { "id": "mixedbread-mxbai-embed-large-v1", "name": "mxbai-embed-large-v1", "level": "model", "family_id": null, "developer": [ "mixedbread" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 335141888, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-03-07", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "English sentence embedding model of BERT-large size, supporting Matryoshka and binary quantised vectors for cheaper retrieval.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/mixedbread-ai/mxbai-embed-large-v1", "https://huggingface.co/api/models?pipeline_tag=feature-extraction&sort=downloads&limit=50", "https://www.modelscope.cn/models/LLM-Research/mxbai-embed-large-v1-gguf" ], "unverified": [] }, { "id": "mixedbread-mxbai-rerank-large-v2", "name": "mxbai-rerank-large-v2", "level": "model", "family_id": null, "developer": [ "mixedbread" ], "model_class": "reranker", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 1543714304, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-03-03", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 109 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Javanese", "Khmer", "Lao", "Malay", "Burmese", "Sundanese", "Thai", "Tagalog", "Vietnamese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Reranker trained with reinforcement learning on a Qwen-2.5 base to re-order search results in many languages.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/mixedbread-ai/mxbai-rerank-large-v2", "https://huggingface.co/api/models?pipeline_tag=text-ranking&sort=downloads&limit=50" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "snowflake-arctic-embed", "name": "Snowflake Arctic Embed", "level": "family", "family_id": null, "developer": [ "snowflake" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-04-11", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Family of open text embedding models for retrieval, from extra small to large, trained by Snowflake.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2405.05374", "https://replicate.com/lucataco/snowflake-arctic-embed-l", "https://catalog.ngc.nvidia.com/orgs/nim/teams/snowflake/models/arctic-embed-l", "https://huggingface.co/Snowflake/snowflake-arctic-embed-m", "https://huggingface.co/Snowflake/snowflake-arctic-embed-l" ], "unverified": [] }, { "id": "snowflake-arctic-embed-m", "name": "snowflake-arctic-embed-m", "level": "variant", "family_id": "snowflake-arctic-embed", "developer": [ "snowflake" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 108891648, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-04-11", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "snowflake-arctic-embed-m, a checkpoint of the Snowflake Arctic Embed family. Family of open text embedding models for retrieval, from extra small to large, trained by Snowflake.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2405.05374", "https://huggingface.co/Snowflake/snowflake-arctic-embed-m", "https://replicate.com/lucataco/snowflake-arctic-embed-l", "https://catalog.ngc.nvidia.com/orgs/nim/teams/snowflake/models/arctic-embed-l" ], "unverified": [] }, { "id": "snowflake-arctic-embed-l", "name": "snowflake-arctic-embed-l", "level": "variant", "family_id": "snowflake-arctic-embed", "developer": [ "snowflake" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 334092288, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-04-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "snowflake-arctic-embed-l, a checkpoint of the Snowflake Arctic Embed family. Family of open text embedding models for retrieval, from extra small to large, trained by Snowflake.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2405.05374", "https://huggingface.co/Snowflake/snowflake-arctic-embed-l", "https://replicate.com/lucataco/snowflake-arctic-embed-l", "https://catalog.ngc.nvidia.com/orgs/nim/teams/snowflake/models/arctic-embed-l" ], "unverified": [] }, { "id": "salesforce-sfr-embedding-mistral", "name": "SFR-Embedding-Mistral", "level": "model", "family_id": null, "developer": [ "salesforce" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 7110660096, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-01-24", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Text embedding model trained from E5-mistral-7b-instruct and Mistral-7B, for retrieval and clustering; research use.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Salesforce/SFR-Embedding-Mistral", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "nvidia-nv-embed-v2", "name": "NV-Embed-v2", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 7851016192, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-08-29", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Generalist text embedding model built on Mistral-7B with a latent-attention pooling layer; non-commercial licence.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/nvidia/NV-Embed-v2", "https://huggingface.co/api/models?pipeline_tag=feature-extraction&sort=likes&limit=50" ], "unverified": [] }, { "id": "nvidia-llama-nemotron-embed-1b-v2", "name": "Llama Nemotron Embed 1B v2", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 1235814400, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-10-16", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "nvidia-open-model-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multilingual text embedding model built on Llama 3.2 1B for question-answer retrieval, offered as an NVIDIA NIM (earlier name NV-EmbedQA 1B v2).", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/nvidia/llama-nemotron-embed-1b-v2", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "nvidia-llama-nemotron-rerank-1b-v2", "name": "Llama Nemotron Rerank 1B v2", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "reranker", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 1235816448, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-10-16", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "openmdw-1-1", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multilingual reranker built on Llama 3.2 1B that scores query and passage pairs for retrieval pipelines (earlier name NV-RerankQA 1B v2).", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/nvidia/llama-nemotron-rerank-1b-v2", "https://huggingface.co/api/models?pipeline_tag=text-ranking&sort=downloads&limit=50", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "naver-splade-cocondenser-selfdistil", "name": "SPLADE (CoCondenser self-distil)", "level": "model", "family_id": null, "developer": [ "naver" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-05-09", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-sa-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Learned sparse retriever that expands queries and documents into weighted vocabulary terms, so it can run on an inverted index.", "verified_on": "2026-10-12", "sources": [ "https://github.com/naver/splade", "https://huggingface.co/naver/splade-cocondenser-selfdistil", "https://huggingface.co/api/models?pipeline_tag=feature-extraction&sort=downloads&limit=50" ], "unverified": [] }, { "id": "allen-ai-specter2", "name": "SPECTER2", "level": "model", "family_id": null, "developer": [ "allen-ai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-02-16", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Scientific paper embedding model trained on citation links, with task adapters, for paper search, recommendation and classification.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/allenai/specter2_base", "https://huggingface.co/api/models?pipeline_tag=feature-extraction&sort=downloads&limit=50" ], "unverified": [] }, { "id": "illuin-colpali", "name": "ColPali", "level": "family", "family_id": null, "developer": [ "illuin-technology" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-06-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Retrieves document pages directly from their images using a vision language model with late-interaction multi-vector embeddings, skipping OCR.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2407.01449", "https://github.com/illuin-tech/colpali", "https://huggingface.co/api/models?pipeline_tag=visual-document-retrieval&sort=downloads&limit=50", "https://www.modelscope.cn/models/AI-ModelScope/colpali", "https://huggingface.co/vidore/colpali-v1.2", "https://huggingface.co/vidore/colqwen2-v1.0" ], "unverified": [] }, { "id": "illuin-colpali-v1-2", "name": "ColPali v1.2", "level": "variant", "family_id": "illuin-colpali", "developer": [ "illuin-technology" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-08-26", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ColPali v1.2, a checkpoint of the ColPali family. Retrieves document pages directly from their images using a vision language model with late-interaction multi-vector embeddings, skipping OCR.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2407.01449", "https://huggingface.co/vidore/colpali-v1.2", "https://huggingface.co/api/models?pipeline_tag=visual-document-retrieval&sort=downloads&limit=50", "https://www.modelscope.cn/models/AI-ModelScope/colpali" ], "unverified": [] }, { "id": "illuin-colqwen2-v1-0", "name": "ColQwen2 v1.0", "level": "variant", "family_id": "illuin-colpali", "developer": [ "illuin-technology" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-11-03", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ColQwen2 v1.0, a checkpoint of the ColPali family. Retrieves document pages directly from their images using a vision language model with late-interaction multi-vector embeddings, skipping OCR.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2407.01449", "https://huggingface.co/vidore/colqwen2-v1.0", "https://huggingface.co/api/models?pipeline_tag=visual-document-retrieval&sort=downloads&limit=50", "https://www.modelscope.cn/models/AI-ModelScope/colpali" ], "unverified": [] }, { "id": "sentence-transformers-all-mpnet-base-v2", "name": "all-mpnet-base-v2", "level": "model", "family_id": null, "developer": [ "sentence-transformers" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 109486978, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server", "laptop" ], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Sentence-transformers model that maps sentences and paragraphs to 768-dimension vectors for clustering and semantic search.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/sentence-transformers/all-mpnet-base-v2", "https://huggingface.co/api/models?pipeline_tag=sentence-similarity&sort=downloads&limit=50", "https://www.modelscope.cn/models/sentence-transformers/all-roberta-large-v1" ], "unverified": [] }, { "id": "sentence-transformers-all-minilm-l12-v2", "name": "all-MiniLM-L12-v2", "level": "model", "family_id": null, "developer": [ "sentence-transformers" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 33360512, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server", "laptop" ], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Sentence-transformers model that maps sentences and paragraphs to 384-dimension vectors for clustering and semantic search.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/sentence-transformers/all-MiniLM-L12-v2", "https://keras.io/keras_hub/presets/", "https://www.kaggle.com/models/keras/sentence-transformers" ], "unverified": [] }, { "id": "sentence-transformers-ms-marco-cross-encoder", "name": "MS MARCO cross-encoders (MiniLM)", "level": "family", "family_id": null, "developer": [ "sentence-transformers" ], "model_class": "reranker", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server", "laptop" ], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small cross-encoders trained on the MS MARCO passage ranking task to re-rank retrieved passages for a query.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/api/models?pipeline_tag=text-ranking&sort=downloads&limit=50", "https://huggingface.co/cross-encoder/ms-marco-MiniLM-L6-v2", "https://huggingface.co/cross-encoder/ms-marco-MiniLM-L12-v2" ], "unverified": [] }, { "id": "sentence-transformers-ms-marco-minilm-l6-v2", "name": "ms-marco-MiniLM-L6-v2", "level": "variant", "family_id": "sentence-transformers-ms-marco-cross-encoder", "developer": [ "sentence-transformers" ], "model_class": "reranker", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 22714113, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ms-marco-MiniLM-L6-v2, a checkpoint of the MS MARCO cross-encoders (MiniLM) family. Small cross-encoders trained on the MS MARCO passage ranking task to re-rank retrieved passages for a query.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/cross-encoder/ms-marco-MiniLM-L6-v2", "https://huggingface.co/api/models?pipeline_tag=text-ranking&sort=downloads&limit=50" ], "unverified": [] }, { "id": "sentence-transformers-ms-marco-minilm-l12-v2", "name": "ms-marco-MiniLM-L12-v2", "level": "variant", "family_id": "sentence-transformers-ms-marco-cross-encoder", "developer": [ "sentence-transformers" ], "model_class": "reranker", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 33360897, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ms-marco-MiniLM-L12-v2, a checkpoint of the MS MARCO cross-encoders (MiniLM) family. Small cross-encoders trained on the MS MARCO passage ranking task to re-rank retrieved passages for a query.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/cross-encoder/ms-marco-MiniLM-L12-v2", "https://huggingface.co/api/models?pipeline_tag=text-ranking&sort=downloads&limit=50" ], "unverified": [] }, { "id": "google-labse", "name": "LaBSE", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 470927360, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2020-07-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "109 languages", "indonesian_support": "listed by developer", "regional_languages": [ "Burmese", "Javanese", "Khmer", "Lao", "Malay", "Sundanese", "Tagalog", "Thai", "Vietnamese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Language-agnostic BERT sentence embeddings for 109 languages, used to find translation pairs and for cross-lingual search.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2007.01852", "https://www.kaggle.com/models/google/labse", "https://huggingface.co/sentence-transformers/LaBSE", "https://huggingface.co/api/models?pipeline_tag=sentence-similarity&sort=likes&limit=50" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "google-universal-sentence-encoder", "name": "Universal Sentence Encoder", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-03-29", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Sentence embedding models, in two variants that trade accuracy against compute, for semantic similarity and transfer to text classification; multilingual, QA, Lite and on-device versions are listed on Kaggle.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1803.11175", "https://www.kaggle.com/models/google/universal-sentence-encoder", "https://www.kaggle.com/models/google/universal-sentence-encoder-qa-ondevice" ], "unverified": [] }, { "id": "google-sentence-t5", "name": "Sentence-T5", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 109628544, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2021-08-19", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Sentence encoders built from the T5 encoder, for sentence similarity and transfer tasks.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2108.08877", "https://www.kaggle.com/models/google/sentence-t5", "https://huggingface.co/sentence-transformers/sentence-t5-base", "https://www.kaggle.com/models/google/gtr" ], "unverified": [] }, { "id": "google-gtr", "name": "GTR (Generalizable T5 Retriever)", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 109628544, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2021-12-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Dual-encoder retrievers built on T5 that generalise to new retrieval domains as the model is scaled up.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2112.07899", "https://www.kaggle.com/models/google/gtr", "https://huggingface.co/sentence-transformers/gtr-t5-base", "https://www.kaggle.com/models/google/sentence-t5" ], "unverified": [] }, { "id": "microsoft-harrier-oss-v1-0-6b", "name": "harrier-oss-v1-0.6b", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 596049920, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2026-03-30", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 94 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Javanese", "Khmer", "Lao", "Malay", "Burmese", "Sundanese", "Thai", "Tagalog", "Vietnamese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Open multilingual text embedding model from Microsoft for retrieval and semantic similarity; a 270M version is also released.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/microsoft/harrier-oss-v1-0.6b", "https://ai.azure.com/catalog/models?task=embeddings", "https://keras.io/keras_hub/presets/" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "tencent-r3-rerank-0-6b", "name": "R3-rerank-0.6b", "level": "model", "family_id": null, "developer": [ "tencent" ], "model_class": "reranker", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 595776512, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2026-07-08", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small Tencent reranker that scores query and document pairs for retrieval pipelines.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/tencent/R3-rerank-0.6b", "https://ai.azure.com/catalog/models?task=rerank" ], "unverified": [] }, { "id": "nih-nlm-medcpt-cross-encoder", "name": "MedCPT Cross-Encoder", "level": "model", "family_id": null, "developer": [ "nih-nlm" ], "model_class": "reranker", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-11-26", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Biomedical cross-encoder from NCBI, trained on PubMed search logs, that re-ranks articles retrieved for a query.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/ncbi/MedCPT-Cross-Encoder", "https://huggingface.co/api/models?pipeline_tag=text-classification&sort=downloads&limit=50" ], "unverified": [] }, { "id": "google-t5", "name": "T5", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Small (60M) to 11B; T5 1.1 retrained versions also released.", "context_length": null, "release_date": "2019-10-23", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, fr, ro, de, multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Encoder-decoder transformer that casts every text task as text-to-text, pre-trained on the C4 corpus, for translation, summarisation and classification.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1910.10683", "https://github.com/google-research/text-to-text-transfer-transformer", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://huggingface.co/api/models?pipeline_tag=translation&sort=downloads&limit=50", "https://huggingface.co/google-t5/t5-small", "https://huggingface.co/google-t5/t5-base" ], "unverified": [] }, { "id": "google-t5-small", "name": "T5 small", "level": "variant", "family_id": "google-t5", "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 60506880, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, fr, ro, de, multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "T5 small, a checkpoint of the T5 family. Encoder-decoder transformer that casts every text task as text-to-text, pre-trained on the C4 corpus, for translation, summarisation and classification.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1910.10683", "https://huggingface.co/google-t5/t5-small", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://huggingface.co/api/models?pipeline_tag=translation&sort=downloads&limit=50" ], "unverified": [] }, { "id": "google-t5-base", "name": "T5 base", "level": "variant", "family_id": "google-t5", "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 222903936, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, fr, ro, de", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "T5 base, a checkpoint of the T5 family. Encoder-decoder transformer that casts every text task as text-to-text, pre-trained on the C4 corpus, for translation, summarisation and classification.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1910.10683", "https://huggingface.co/google-t5/t5-base", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://huggingface.co/api/models?pipeline_tag=translation&sort=downloads&limit=50" ], "unverified": [] }, { "id": "google-t5-large", "name": "T5 large", "level": "variant", "family_id": "google-t5", "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 737668608, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, fr, ro, de, multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "T5 large, a checkpoint of the T5 family. Encoder-decoder transformer that casts every text task as text-to-text, pre-trained on the C4 corpus, for translation, summarisation and classification.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1910.10683", "https://huggingface.co/google-t5/t5-large", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://huggingface.co/api/models?pipeline_tag=translation&sort=downloads&limit=50" ], "unverified": [] }, { "id": "google-t5-11b", "name": "T5 11B", "level": "variant", "family_id": "google-t5", "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, fr, ro, de, multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "T5 11B, a checkpoint of the T5 family. Encoder-decoder transformer that casts every text task as text-to-text, pre-trained on the C4 corpus, for translation, summarisation and classification.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1910.10683", "https://huggingface.co/google-t5/t5-11b", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://huggingface.co/api/models?pipeline_tag=translation&sort=downloads&limit=50" ], "unverified": [] }, { "id": "google-flan-t5", "name": "Flan-T5", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-10-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, fr, ro, de, multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "T5 checkpoints instruction-tuned on a large task collection (1.8K tasks in the paper), giving better zero-shot and few-shot behaviour for question answering and summarisation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2210.11416", "https://cloud.google.com/model-garden", "https://huggingface.co/api/models?pipeline_tag=question-answering&sort=downloads&limit=50", "https://huggingface.co/google/flan-t5-base", "https://huggingface.co/google/flan-t5-large" ], "unverified": [] }, { "id": "google-flan-t5-base", "name": "Flan-T5 base", "level": "variant", "family_id": "google-flan-t5", "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 247577856, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2022-10-21", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, fr, ro, de, multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Flan-T5 base, a checkpoint of the Flan-T5 family. T5 checkpoints instruction-tuned on a large task collection (1.8K tasks in the paper), giving better zero-shot and few-shot behaviour for question answering and summarisation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2210.11416", "https://huggingface.co/google/flan-t5-base", "https://cloud.google.com/model-garden", "https://huggingface.co/api/models?pipeline_tag=question-answering&sort=downloads&limit=50" ], "unverified": [] }, { "id": "google-flan-t5-large", "name": "Flan-T5 large", "level": "variant", "family_id": "google-flan-t5", "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 783150080, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2022-10-21", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, fr, ro, de, multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Flan-T5 large, a checkpoint of the Flan-T5 family. T5 checkpoints instruction-tuned on a large task collection (1.8K tasks in the paper), giving better zero-shot and few-shot behaviour for question answering and summarisation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2210.11416", "https://huggingface.co/google/flan-t5-large", "https://cloud.google.com/model-garden", "https://huggingface.co/api/models?pipeline_tag=question-answering&sort=downloads&limit=50" ], "unverified": [] }, { "id": "google-flan-t5-xxl", "name": "Flan-T5 XXL", "level": "variant", "family_id": "google-flan-t5", "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 11266928640, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2022-10-21", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, fr, ro, de, multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Flan-T5 XXL, a checkpoint of the Flan-T5 family. T5 checkpoints instruction-tuned on a large task collection (1.8K tasks in the paper), giving better zero-shot and few-shot behaviour for question answering and summarisation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2210.11416", "https://huggingface.co/google/flan-t5-xxl", "https://cloud.google.com/model-garden", "https://huggingface.co/api/models?pipeline_tag=question-answering&sort=downloads&limit=50" ], "unverified": [] }, { "id": "google-umt5", "name": "UMT5", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-04-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 102 languages", "indonesian_support": "not stated", "regional_languages": [ "Javanese", "Khmer", "Lao", "Malay", "Burmese", "Sundanese", "Thai", "Vietnamese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Multilingual T5 retrained with the UniMax language sampling method over 107 languages, an updated mT5.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2304.09151", "https://huggingface.co/google/umt5-xxl", "https://github.com/onnx/models/tree/main/Natural_Language_Processing/umt5encoder_Opset16_transformers" ], "unverified": [] }, { "id": "google-longt5", "name": "LongT5", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-12-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "T5 variant with local and transient-global attention for long inputs, used for long-document summarisation and question answering.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2112.07916", "https://huggingface.co/google/long-t5-tglobal-base", "https://huggingface.co/api/models?pipeline_tag=summarization&sort=downloads&limit=50", "https://github.com/onnx/models/tree/main/Natural_Language_Processing/longt5encoder_Opset16_transformers" ], "unverified": [] }, { "id": "meta-bart", "name": "BART", "level": "family", "family_id": null, "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-10-29", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Denoising sequence-to-sequence pre-training that corrupts text and learns to rebuild it, strong for summarisation; the MNLI fine-tune is widely used for zero-shot classification.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1910.13461", "https://github.com/facebookresearch/fairseq/tree/main/examples/bart", "https://www.kaggle.com/models/keras/bart", "https://paperswithcode.com/method/bart", "https://huggingface.co/facebook/bart-large", "https://huggingface.co/facebook/bart-large-cnn" ], "unverified": [] }, { "id": "meta-bart-large", "name": "BART large", "level": "variant", "family_id": "meta-bart", "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BART large, a checkpoint of the BART family. Denoising sequence-to-sequence pre-training that corrupts text and learns to rebuild it, strong for summarisation; the MNLI fine-tune is widely used for zero-shot classification.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1910.13461", "https://huggingface.co/facebook/bart-large", "https://www.kaggle.com/models/keras/bart", "https://paperswithcode.com/method/bart" ], "unverified": [] }, { "id": "meta-bart-large-cnn", "name": "BART large CNN", "level": "variant", "family_id": "meta-bart", "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 406290432, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BART large fine-tuned on CNN/Daily Mail for English news summarisation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1910.13461", "https://huggingface.co/facebook/bart-large-cnn", "https://www.kaggle.com/models/keras/bart", "https://paperswithcode.com/method/bart" ], "unverified": [] }, { "id": "meta-bart-large-mnli", "name": "BART large MNLI", "level": "variant", "family_id": "meta-bart", "developer": [ "meta" ], "model_class": "language encoder", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 407344133, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BART large fine-tuned on MultiNLI, used as a zero-shot text classifier by framing labels as hypotheses.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1910.13461", "https://huggingface.co/facebook/bart-large-mnli", "https://www.kaggle.com/models/keras/bart", "https://paperswithcode.com/method/bart" ], "unverified": [] }, { "id": "hugging-face-distilbart-cnn-12-6", "name": "DistilBART CNN 12-6", "level": "variant", "family_id": "meta-bart", "developer": [ "hugging-face" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-10-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Smaller BART for news summarisation, distilled from BART large CNN by shrinking the decoder.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2010.13002", "https://huggingface.co/sshleifer/distilbart-cnn-12-6", "https://www.kaggle.com/models/keras/bart", "https://paperswithcode.com/method/bart" ], "unverified": [] }, { "id": "meta-mbart-large-50-many-to-many-mmt", "name": "mBART-50 many-to-many", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "translation", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 611129542, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 53 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Burmese", "Vietnamese", "Khmer", "Thai", "Tagalog" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Multilingual BART fine-tuned to translate directly between any pair of 50 languages, including Indonesian.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt", "https://huggingface.co/api/models?pipeline_tag=translation&sort=downloads&limit=50" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "google-pegasus", "name": "PEGASUS", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-12-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Encoder-decoder pre-trained by removing whole sentences and generating them, built for abstractive summarisation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1912.08777", "https://github.com/google-research/pegasus", "https://paperswithcode.com/method/pegasus", "https://github.com/onnx/models/tree/main/Natural_Language_Processing/pegasus_Opset16_transformers", "https://huggingface.co/google/pegasus-large", "https://huggingface.co/google/pegasus-xsum" ], "unverified": [] }, { "id": "google-pegasus-large", "name": "PEGASUS large", "level": "variant", "family_id": "google-pegasus", "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "PEGASUS large, a checkpoint of the PEGASUS family. Encoder-decoder pre-trained by removing whole sentences and generating them, built for abstractive summarisation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1912.08777", "https://huggingface.co/google/pegasus-large", "https://paperswithcode.com/method/pegasus", "https://github.com/onnx/models/tree/main/Natural_Language_Processing/pegasus_Opset16_transformers" ], "unverified": [] }, { "id": "google-pegasus-xsum", "name": "PEGASUS XSum", "level": "variant", "family_id": "google-pegasus", "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "PEGASUS XSum, a checkpoint of the PEGASUS family. Encoder-decoder pre-trained by removing whole sentences and generating them, built for abstractive summarisation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1912.08777", "https://huggingface.co/google/pegasus-xsum", "https://paperswithcode.com/method/pegasus", "https://github.com/onnx/models/tree/main/Natural_Language_Processing/pegasus_Opset16_transformers" ], "unverified": [] }, { "id": "google-bigbird-pegasus-large-arxiv", "name": "BigBird-PEGASUS large arXiv", "level": "variant", "family_id": "google-pegasus", "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-07-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "PEGASUS with BigBird sparse attention, fine-tuned to summarise long scientific papers.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2007.14062", "https://huggingface.co/google/bigbird-pegasus-large-arxiv", "https://paperswithcode.com/method/pegasus", "https://github.com/onnx/models/tree/main/Natural_Language_Processing/pegasus_Opset16_transformers" ], "unverified": [] }, { "id": "openai-gpt-2", "name": "GPT-2", "level": "family", "family_id": null, "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Four sizes (small, medium, large, XL); parameter counts are on each variant record.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Decoder-only language model from the 2019 paper Language Models are Unsupervised Multitask Learners, trained on WebText to predict the next word and released in stages.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openai/gpt-2", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/gpt-2", "https://huggingface.co/api/models?pipeline_tag=text-generation&sort=downloads&limit=50", "https://huggingface.co/openai-community/gpt2", "https://huggingface.co/openai-community/gpt2-medium" ], "unverified": [] }, { "id": "openai-gpt2", "name": "GPT-2 (124M)", "level": "variant", "family_id": "openai-gpt-2", "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 137022720, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "GPT-2 (124M), a checkpoint of the GPT-2 family. Decoder-only language model from the 2019 paper Language Models are Unsupervised Multitask Learners, trained on WebText to predict the next word and released in stages.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/openai-community/gpt2", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/gpt-2", "https://huggingface.co/api/models?pipeline_tag=text-generation&sort=downloads&limit=50" ], "unverified": [] }, { "id": "openai-gpt2-medium", "name": "GPT-2 medium", "level": "variant", "family_id": "openai-gpt-2", "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 379988992, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "GPT-2 medium, a checkpoint of the GPT-2 family. Decoder-only language model from the 2019 paper Language Models are Unsupervised Multitask Learners, trained on WebText to predict the next word and released in stages.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/openai-community/gpt2-medium", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/gpt-2", "https://huggingface.co/api/models?pipeline_tag=text-generation&sort=downloads&limit=50" ], "unverified": [] }, { "id": "openai-gpt2-large", "name": "GPT-2 large", "level": "variant", "family_id": "openai-gpt-2", "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 811778816, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "GPT-2 large, a checkpoint of the GPT-2 family. Decoder-only language model from the 2019 paper Language Models are Unsupervised Multitask Learners, trained on WebText to predict the next word and released in stages.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/openai-community/gpt2-large", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/gpt-2", "https://huggingface.co/api/models?pipeline_tag=text-generation&sort=downloads&limit=50" ], "unverified": [] }, { "id": "openai-gpt2-xl", "name": "GPT-2 XL", "level": "variant", "family_id": "openai-gpt-2", "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 1607942848, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "GPT-2 XL, a checkpoint of the GPT-2 family. Decoder-only language model from the 2019 paper Language Models are Unsupervised Multitask Learners, trained on WebText to predict the next word and released in stages.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/openai-community/gpt2-xl", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/gpt-2", "https://huggingface.co/api/models?pipeline_tag=text-generation&sort=downloads&limit=50" ], "unverified": [] }, { "id": "openai-gpt-1", "name": "GPT (GPT-1)", "level": "model", "family_id": null, "developer": [ "openai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 119680512, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "The first OpenAI generative pre-trained transformer, a 12-layer decoder trained on books and then fine-tuned for language understanding tasks.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/openai-community/openai-gpt", "https://pytorch.org/hub/huggingface_pytorch-transformers" ], "unverified": [] }, { "id": "eleutherai-gpt-j-6b", "name": "GPT-J 6B", "level": "model", "family_id": null, "developer": [ "eleutherai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open 6-billion-parameter English language model trained on the Pile with Mesh Transformer JAX.", "verified_on": "2026-10-12", "sources": [ "https://github.com/kingoflolz/mesh-transformer-jax", "https://huggingface.co/EleutherAI/gpt-j-6b", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html", "https://huggingface.co/OpenVINO/gpt-j-6b-fp16-ov" ], "unverified": [] }, { "id": "eleutherai-gpt-neox-20b", "name": "GPT-NeoX-20B", "level": "model", "family_id": null, "developer": [ "eleutherai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 20739117584, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2022-02-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open 20-billion-parameter autoregressive language model trained on the Pile, released with its weights and training code.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2204.06745", "https://huggingface.co/EleutherAI/gpt-neox-20b", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html", "https://github.com/onnx/models/tree/main/Generative_AI/gptneox_Opset16_transformers", "https://blog.eleuther.ai/announcing-20b/" ], "unverified": [] }, { "id": "eleutherai-gpt-neo", "name": "GPT-Neo", "level": "family", "family_id": null, "developer": [ "eleutherai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "EleutherAI's first open GPT-3-style models trained on the Pile with Mesh TensorFlow.", "verified_on": "2026-10-12", "sources": [ "https://github.com/EleutherAI/gpt-neo", "https://paperswithcode.com/method/gpt-neo", "https://huggingface.co/EleutherAI/gpt-neo-1.3B", "https://huggingface.co/EleutherAI/gpt-neo-2.7B" ], "unverified": [] }, { "id": "eleutherai-gpt-neo-1-3b", "name": "GPT-Neo 1.3B", "level": "variant", "family_id": "eleutherai-gpt-neo", "developer": [ "eleutherai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 1365907456, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "GPT-Neo 1.3B, a checkpoint of the GPT-Neo family. EleutherAI's first open GPT-3-style models trained on the Pile with Mesh TensorFlow.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/EleutherAI/gpt-neo-1.3B", "https://paperswithcode.com/method/gpt-neo" ], "unverified": [] }, { "id": "eleutherai-gpt-neo-2-7b", "name": "GPT-Neo 2.7B", "level": "variant", "family_id": "eleutherai-gpt-neo", "developer": [ "eleutherai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 2718416384, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "GPT-Neo 2.7B, a checkpoint of the GPT-Neo family. EleutherAI's first open GPT-3-style models trained on the Pile with Mesh TensorFlow.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/EleutherAI/gpt-neo-2.7B", "https://paperswithcode.com/method/gpt-neo" ], "unverified": [] }, { "id": "bigscience-bloom", "name": "BLOOM", "level": "family", "family_id": null, "developer": [ "bigscience" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "560M to 176B; BLOOMZ is the multitask fine-tuned version.", "context_length": null, "release_date": "2022-05-19", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "bigscience-bloom-rail-1-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "46 natural languages including Indonesian, and 13 programming languages", "indonesian_support": "listed by developer", "regional_languages": [ "Vietnamese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Open multilingual language model built by the BigScience collaboration of hundreds of researchers, trained on 46 natural and 13 programming languages including Indonesian.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2211.05100", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html", "https://huggingface.co/api/models?pipeline_tag=text-generation&sort=likes&limit=50", "https://huggingface.co/bigscience/bloom", "https://huggingface.co/bigscience/bloom-560m" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "bigscience-bloom-176b", "name": "BLOOM 176B", "level": "variant", "family_id": "bigscience-bloom", "developer": [ "bigscience" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 176247271424, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2022-05-19", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "bigscience-bloom-rail-1-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 46 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Vietnamese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "BLOOM 176B, a checkpoint of the BLOOM family. Open multilingual language model built by the BigScience collaboration of hundreds of researchers, trained on 46 natural and 13 programming languages including Indonesian.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2211.05100", "https://huggingface.co/bigscience/bloom", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html", "https://huggingface.co/api/models?pipeline_tag=text-generation&sort=likes&limit=50" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "bigscience-bloom-560m", "name": "BLOOM 560M", "level": "variant", "family_id": "bigscience-bloom", "developer": [ "bigscience" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 559214592, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2022-05-19", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "bigscience-bloom-rail-1-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 48 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Vietnamese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "BLOOM 560M, a checkpoint of the BLOOM family. Open multilingual language model built by the BigScience collaboration of hundreds of researchers, trained on 46 natural and 13 programming languages including Indonesian.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2211.05100", "https://huggingface.co/bigscience/bloom-560m", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html", "https://huggingface.co/api/models?pipeline_tag=text-generation&sort=likes&limit=50" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "bigscience-bloom-7b1", "name": "BLOOM 7B1", "level": "variant", "family_id": "bigscience-bloom", "developer": [ "bigscience" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 7069016064, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2022-05-19", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "bigscience-bloom-rail-1-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 48 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Vietnamese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "BLOOM 7B1, a checkpoint of the BLOOM family. Open multilingual language model built by the BigScience collaboration of hundreds of researchers, trained on 46 natural and 13 programming languages including Indonesian.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2211.05100", "https://huggingface.co/bigscience/bloom-7b1", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html", "https://huggingface.co/api/models?pipeline_tag=text-generation&sort=likes&limit=50" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "bigscience-bloomz", "name": "BLOOMZ", "level": "variant", "family_id": "bigscience-bloom", "developer": [ "bigscience" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 176247271424, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2022-11-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "bigscience-bloom-rail-1-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 46 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Vietnamese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "BLOOM fine-tuned on the xP3 multilingual multitask mixture so it follows prompts in dozens of languages zero-shot.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2211.01786", "https://huggingface.co/bigscience/bloomz", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html", "https://huggingface.co/api/models?pipeline_tag=text-generation&sort=likes&limit=50" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "bigscience-bloomz-7b1", "name": "BLOOMZ 7B1", "level": "variant", "family_id": "bigscience-bloom", "developer": [ "bigscience" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 7069016064, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2022-11-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "bigscience-bloom-rail-1-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 46 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Vietnamese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "BLOOM 7B1 fine-tuned on the xP3 multilingual multitask mixture to follow prompts zero-shot.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2211.01786", "https://huggingface.co/bigscience/bloomz-7b1", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html", "https://huggingface.co/api/models?pipeline_tag=text-generation&sort=likes&limit=50" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "meta-opt", "name": "OPT", "level": "family", "family_id": null, "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-05-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open Pre-trained Transformers, decoder-only models from 125M to 175B parameters released to let researchers study large language models.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2205.01068", "https://github.com/facebookresearch/metaseq", "https://huggingface.co/api/models?pipeline_tag=text-generation&sort=downloads&limit=50", "https://www.kaggle.com/models/keras/opt", "https://huggingface.co/facebook/opt-125m", "https://huggingface.co/facebook/opt-1.3b" ], "unverified": [] }, { "id": "meta-opt-125m", "name": "OPT 125M", "level": "variant", "family_id": "meta-opt", "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-05-11", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OPT 125M, a checkpoint of the OPT family. Open Pre-trained Transformers, decoder-only models from 125M to 175B parameters released to let researchers study large language models.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2205.01068", "https://huggingface.co/facebook/opt-125m", "https://huggingface.co/api/models?pipeline_tag=text-generation&sort=downloads&limit=50", "https://www.kaggle.com/models/keras/opt" ], "unverified": [] }, { "id": "meta-opt-1-3b", "name": "OPT 1.3B", "level": "variant", "family_id": "meta-opt", "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-05-11", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OPT 1.3B, a checkpoint of the OPT family. Open Pre-trained Transformers, decoder-only models from 125M to 175B parameters released to let researchers study large language models.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2205.01068", "https://huggingface.co/facebook/opt-1.3b", "https://huggingface.co/api/models?pipeline_tag=text-generation&sort=downloads&limit=50", "https://www.kaggle.com/models/keras/opt" ], "unverified": [] }, { "id": "meta-opt-6-7b", "name": "OPT 6.7B", "level": "variant", "family_id": "meta-opt", "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-05-11", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OPT 6.7B, a checkpoint of the OPT family. Open Pre-trained Transformers, decoder-only models from 125M to 175B parameters released to let researchers study large language models.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2205.01068", "https://huggingface.co/facebook/opt-6.7b", "https://huggingface.co/api/models?pipeline_tag=text-generation&sort=downloads&limit=50", "https://www.kaggle.com/models/keras/opt" ], "unverified": [] }, { "id": "meta-opt-66b", "name": "OPT 66B", "level": "variant", "family_id": "meta-opt", "developer": [ "meta" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-06-23", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OPT 66B, a checkpoint of the OPT family. Open Pre-trained Transformers, decoder-only models from 125M to 175B parameters released to let researchers study large language models.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2205.01068", "https://huggingface.co/facebook/opt-66b", "https://huggingface.co/api/models?pipeline_tag=text-generation&sort=downloads&limit=50", "https://www.kaggle.com/models/keras/opt" ], "unverified": [] }, { "id": "tii-falcon", "name": "Falcon (first generation)", "level": "family", "family_id": null, "developer": [ "tii" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-04-24", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, de, es, fr", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "TII's first open Falcon decoder models, trained mostly on the RefinedWeb web corpus, from 1B to 180B parameters.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2311.16867", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://huggingface.co/tiiuae/falcon-7b", "https://huggingface.co/tiiuae/falcon-7b-instruct" ], "unverified": [] }, { "id": "tii-falcon-7b", "name": "Falcon-7B", "level": "variant", "family_id": "tii-falcon", "developer": [ "tii" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 7217189760, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-04-24", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Falcon-7B, a checkpoint of the Falcon (first generation) family. TII's first open Falcon decoder models, trained mostly on the RefinedWeb web corpus, from 1B to 180B parameters.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2311.16867", "https://huggingface.co/tiiuae/falcon-7b", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "tii-falcon-7b-instruct", "name": "Falcon-7B-Instruct", "level": "variant", "family_id": "tii-falcon", "developer": [ "tii" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 7217189760, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-04-25", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Falcon-7B-Instruct, a checkpoint of the Falcon (first generation) family. TII's first open Falcon decoder models, trained mostly on the RefinedWeb web corpus, from 1B to 180B parameters.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2311.16867", "https://huggingface.co/tiiuae/falcon-7b-instruct", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "tii-falcon-40b", "name": "Falcon-40B", "level": "variant", "family_id": "tii-falcon", "developer": [ "tii" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 41835970560, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-05-24", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, de, es, fr", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Falcon-40B, a checkpoint of the Falcon (first generation) family. TII's first open Falcon decoder models, trained mostly on the RefinedWeb web corpus, from 1B to 180B parameters.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2311.16867", "https://huggingface.co/tiiuae/falcon-40b", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "tii-falcon-180b", "name": "Falcon-180B", "level": "variant", "family_id": "tii-falcon", "developer": [ "tii" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 179522565120, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-08-28", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, de, es, fr", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Falcon-180B, a checkpoint of the Falcon (first generation) family. TII's first open Falcon decoder models, trained mostly on the RefinedWeb web corpus, from 1B to 180B parameters.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2311.16867", "https://huggingface.co/tiiuae/falcon-180B", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "tii-falcon-rw-1b", "name": "Falcon-RW-1B", "level": "variant", "family_id": "tii-falcon", "developer": [ "tii" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-06-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Falcon-RW-1B, a checkpoint of the Falcon (first generation) family. TII's first open Falcon decoder models, trained mostly on the RefinedWeb web corpus, from 1B to 180B parameters.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2306.01116", "https://huggingface.co/tiiuae/falcon-rw-1b", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "tii-falcon-11b", "name": "Falcon 2 11B", "level": "model", "family_id": null, "developer": [ "tii" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 11102822400, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-05-09", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, de, es, fr, it, nl, pl, pt, ro, cs", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Second-generation 11-billion-parameter Falcon model trained on over 5 trillion tokens of RefinedWeb and curated data, multilingual.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/tiiuae/falcon-11B", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "tii-falcon3-7b-instruct", "name": "Falcon3-7B-Instruct", "level": "model", "family_id": null, "developer": [ "tii" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 7455550464, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-11-29", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "falcon-llm-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, fr, es, pt", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Instruction-tuned 7B model from TII's Falcon3 series, for reasoning, coding and chat in English, French, Spanish and Portuguese.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/tiiuae/Falcon3-7B-Instruct", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html", "https://aihub.qualcomm.com/models/falcon_v3_7b_instruct" ], "unverified": [] }, { "id": "mistral-ai-mixtral-8x7b-v0-1", "name": "Mixtral 8x7B v0.1", "level": "model", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 46702792704, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-12-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: fr, it, de, es, en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Sparse mixture-of-experts language model with 8 experts per layer, of which 2 are used per token, released under Apache 2.0.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2401.04088", "https://huggingface.co/mistralai/Mixtral-8x7B-v0.1", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://keras.io/keras_hub/presets/", "https://mistral.ai/news/mixtral-of-experts" ], "unverified": [] }, { "id": "mistral-ai-mixtral-8x7b-instruct-v0-1", "name": "Mixtral 8x7B Instruct v0.1", "level": "model", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 46702792704, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-12-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: fr, it, de, es, en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Instruction-tuned version of Mixtral 8x7B for chat and following instructions.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2401.04088", "https://huggingface.co/mistralai/Mixtral-8x7B-Instruct-v0.1", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://keras.io/keras_hub/presets/", "https://mistral.ai/news/mixtral-of-experts" ], "unverified": [] }, { "id": "mistral-ai-mistral-nemo-instruct-2407", "name": "Mistral NeMo Instruct 2407", "level": "model", "family_id": null, "developer": [ "mistral-ai", "nvidia" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 12247782400, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-07-17", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, fr, de, es, it, pt, ru, zh, ja", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "12-billion-parameter instruction-tuned model built jointly by Mistral AI and NVIDIA, with a 128k context window, released under Apache 2.0.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/mistralai/Mistral-Nemo-Instruct-2407", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html", "https://catalog.ngc.nvidia.com/orgs/nvidia/models/mistral-nemo-12b-instruct-onnx-int4-rtx" ], "unverified": [] }, { "id": "mistral-ai-ministral-3-14b-instruct-2512", "name": "Ministral 3 14B Instruct (25.12)", "level": "model", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 13945032240, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-10-31", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, fr, es, de, it, pt, nl, zh, ja, ko, ar", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Largest model of the Ministral 3 series (December 2025), an open Apache 2.0 model with text and vision capabilities; the 3B and 8B members are separate AURION records.", "verified_on": "2026-10-12", "sources": [ "https://docs.mistral.ai/getting-started/models/models_overview/", "https://huggingface.co/mistralai/Ministral-3-14B-Instruct-2512", "https://docs.aws.amazon.com/bedrock/latest/userguide/model-cards.html" ], "unverified": [] }, { "id": "nvidia-mistral-nemo-minitron-8b-base", "name": "Mistral-NeMo-Minitron 8B base", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8414105600, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-08-19", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "nvidia-open-model-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Base model made by pruning and distilling Mistral NeMo 12B down to 8 billion parameters.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/nvidia/Mistral-NeMo-Minitron-8B-Base", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/ace/models/mistral-nemo-minitron-2b-128k-instruct" ], "unverified": [] }, { "id": "hugging-face-zephyr-7b-beta", "name": "Zephyr 7B beta", "level": "model", "family_id": null, "developer": [ "hugging-face" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 7241732096, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-10-26", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Chat model fine-tuned from Mistral-7B with distilled supervised fine-tuning and direct preference optimisation on synthetic data.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/HuggingFaceH4/zephyr-7b-beta", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html", "https://huggingface.co/OpenVINO/zephyr-7b-beta-int4-ov" ], "unverified": [] }, { "id": "allen-ai-olmo-7b", "name": "OLMo 7B (first release)", "level": "model", "family_id": null, "developer": [ "allen-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 6888095744, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-01-09", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First fully open OLMo language model, released with its training data, code, logs and intermediate checkpoints for research on language models.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/allenai/OLMo-7B", "https://www.kaggle.com/models/allenai/olmo" ], "unverified": [] }, { "id": "allen-ai-tulu", "name": "Tulu", "level": "family", "family_id": null, "developer": [ "allen-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-11-12", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Ai2 post-training recipes and models that turn open base models (Llama) into instruction followers, with all data and code released.", "verified_on": "2026-10-12", "sources": [ "https://www.kaggle.com/models/allenai/tulu-2", "https://www.kaggle.com/models/allenai/tulu-3", "https://huggingface.co/allenai/tulu-2-dpo-70b", "https://huggingface.co/allenai/Llama-3.1-Tulu-3-8B" ], "unverified": [] }, { "id": "allen-ai-tulu-2-dpo-70b", "name": "Tulu 2 DPO 70B", "level": "variant", "family_id": "allen-ai-tulu", "developer": [ "allen-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 68976648192, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-11-12", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tulu 2 DPO 70B, a checkpoint of the Tulu family. Ai2 post-training recipes and models that turn open base models (Llama) into instruction followers, with all data and code released.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/allenai/tulu-2-dpo-70b", "https://www.kaggle.com/models/allenai/tulu-2", "https://www.kaggle.com/models/allenai/tulu-3" ], "unverified": [] }, { "id": "allen-ai-llama-3-1-tulu-3-8b", "name": "Llama 3.1 Tulu 3 8B", "level": "variant", "family_id": "allen-ai-tulu", "developer": [ "allen-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8030326784, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-11-20", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "llama-3-1-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Llama 3.1 Tulu 3 8B, a checkpoint of the Tulu family. Ai2 post-training recipes and models that turn open base models (Llama) into instruction followers, with all data and code released.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/allenai/Llama-3.1-Tulu-3-8B", "https://www.kaggle.com/models/allenai/tulu-2", "https://www.kaggle.com/models/allenai/tulu-3" ], "unverified": [] }, { "id": "allen-ai-llama-3-1-tulu-3-70b", "name": "Llama 3.1 Tulu 3 70B", "level": "variant", "family_id": "allen-ai-tulu", "developer": [ "allen-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 70553837568, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-11-20", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "llama-3-1-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Llama 3.1 Tulu 3 70B, a checkpoint of the Tulu family. Ai2 post-training recipes and models that turn open base models (Llama) into instruction followers, with all data and code released.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/allenai/Llama-3.1-Tulu-3-70B", "https://www.kaggle.com/models/allenai/tulu-2", "https://www.kaggle.com/models/allenai/tulu-3" ], "unverified": [] }, { "id": "intel-neural-chat-7b-v3-3", "name": "Neural-Chat 7B v3.3", "level": "model", "family_id": null, "developer": [ "intel" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 7241732096, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-12-09", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Mistral-7B fine-tuned by Intel on Gaudi 2 accelerators with supervised and preference data, for chat.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Intel/neural-chat-7b-v3-3", "https://huggingface.co/OpenVINO/neural-chat-7b-v3-3-fp16-ov" ], "unverified": [] }, { "id": "adept-persimmon-8b-chat", "name": "Persimmon-8B-Chat", "level": "model", "family_id": null, "developer": [ "adept" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-09-07", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open 8-billion-parameter chat model from Adept with a 16k context, released under Apache 2.0.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/adept/persimmon-8b-chat", "https://huggingface.co/OpenVINO/persimmon-8b-chat-int4-ov" ], "unverified": [] }, { "id": "microsoft-phi-2", "name": "Phi-2", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 2779683840, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-12-13", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU", "laptop" ], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small 2.7-billion-parameter language model trained on synthetic textbook-style and filtered web data, for research on small models.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/microsoft/phi-2", "https://huggingface.co/OpenVINO/phi-2-int8-ov", "https://ai.google.dev/edge/mediapipe/solutions/genai/llm_inference" ], "unverified": [] }, { "id": "microsoft-phi-3-5-vision-instruct", "name": "Phi-3.5-vision-instruct", "level": "model", "family_id": "microsoft-phi-3", "developer": [ "microsoft" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 4146621440, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-08-16", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Lightweight open multimodal model from the Phi-3 family that reads text and images, including multiple frames, for document and chart understanding.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/microsoft/Phi-3.5-vision-instruct", "https://huggingface.co/OpenVINO/Phi-3.5-vision-instruct-fp16-ov" ], "unverified": [] }, { "id": "microsoft-biogpt", "name": "BioGPT", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-11-20", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Domain-specific generative transformer pre-trained on PubMed literature for biomedical text generation and mining.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/microsoft/biogpt", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://ai.azure.com/catalog/models?industry=health-and-life-sciences" ], "unverified": [] }, { "id": "stanford-university-biomedlm", "name": "BioMedLM", "level": "model", "family_id": null, "developer": [ "stanford-university" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-12-14", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "bigscience-bloom-rail-1-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "2.7-billion-parameter GPT-style model trained only on PubMed abstracts and papers, for biomedical question answering research.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/stanford-crfm/BioMedLM", "https://ai.azure.com/catalog/models?industry=health-and-life-sciences" ], "unverified": [] }, { "id": "biomistral-biomistral-7b", "name": "BioMistral 7B", "level": "model", "family_id": null, "developer": [ "biomistral" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-02-14", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: fr, en, de, nl, es, pt, pl, ro, it", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Mistral-7B further pre-trained on PubMed Central for medical question answering; released for research, not for clinical use.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/BioMistral/BioMistral-7B", "https://ai.azure.com/catalog/models?industry=health-and-life-sciences" ], "unverified": [] }, { "id": "hpai-bsc-llama3-1-aloe-beta-8b", "name": "Aloe Beta 8B", "level": "model", "family_id": null, "developer": [ "barcelona-supercomputing-center" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8030261248, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-10-30", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open healthcare language model fine-tuned from Llama 3.1 8B on medical and general data, with safety alignment.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/HPAI-BSC/Llama3.1-Aloe-Beta-8B", "https://huggingface.co/api/models?pipeline_tag=question-answering&sort=downloads&limit=50", "https://www.kaggle.com/models/hpai-bsc/llama3.1-aloe-beta" ], "unverified": [] }, { "id": "amazon-alexatm-20b", "name": "AlexaTM 20B", "level": "model", "family_id": null, "developer": [ "amazon" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-08-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multilingual 20-billion-parameter sequence-to-sequence model pre-trained on denoising and causal language modelling, strong at few-shot summarisation and translation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2208.01448", "https://github.com/amazon-science/alexa-teacher-models", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "snowflake-arctic-instruct", "name": "Snowflake Arctic Instruct", "level": "model", "family_id": null, "developer": [ "snowflake" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 478584608768, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-04-21", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Dense-MoE hybrid model with 480 billion total and 17 billion active parameters, aimed at enterprise SQL, coding and instruction following.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Snowflake/snowflake-arctic-instruct", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html", "https://replicate.com/snowflake/snowflake-arctic-instruct" ], "unverified": [] }, { "id": "meta-code-llama", "name": "Code Llama", "level": "family", "family_id": null, "developer": [ "meta" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-08-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "llama-2-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: code", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Llama 2 further trained on code, with base, Python and instruction versions from 7B to 70B, for code completion, infilling and chat about code.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2308.12950", "https://github.com/meta-llama/codellama", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://www.kaggle.com/models/metaresearch/codellama", "https://huggingface.co/meta-llama/CodeLlama-7b-hf", "https://huggingface.co/meta-llama/CodeLlama-13b-Instruct-hf" ], "unverified": [] }, { "id": "meta-codellama-7b", "name": "Code Llama 7B", "level": "variant", "family_id": "meta-code-llama", "developer": [ "meta" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 6738546688, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights restricted", "licence": "llama-2-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: code", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Code Llama 7B, a checkpoint of the Code Llama family. Llama 2 further trained on code, with base, Python and instruction versions from 7B to 70B, for code completion, infilling and chat about code.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2308.12950", "https://huggingface.co/meta-llama/CodeLlama-7b-hf", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://www.kaggle.com/models/metaresearch/codellama" ], "unverified": [] }, { "id": "meta-codellama-13b-instruct", "name": "Code Llama 13B Instruct", "level": "variant", "family_id": "meta-code-llama", "developer": [ "meta" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 13016028160, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights restricted", "licence": "llama-2-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: code", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Code Llama 13B Instruct, a checkpoint of the Code Llama family. Llama 2 further trained on code, with base, Python and instruction versions from 7B to 70B, for code completion, infilling and chat about code.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2308.12950", "https://huggingface.co/meta-llama/CodeLlama-13b-Instruct-hf", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://www.kaggle.com/models/metaresearch/codellama" ], "unverified": [] }, { "id": "meta-codellama-34b-python", "name": "Code Llama 34B Python", "level": "variant", "family_id": "meta-code-llama", "developer": [ "meta" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 33743970304, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights restricted", "licence": "llama-2-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: code", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Code Llama 34B Python, a checkpoint of the Code Llama family. Llama 2 further trained on code, with base, Python and instruction versions from 7B to 70B, for code completion, infilling and chat about code.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2308.12950", "https://huggingface.co/meta-llama/CodeLlama-34b-Python-hf", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://www.kaggle.com/models/metaresearch/codellama" ], "unverified": [] }, { "id": "meta-codellama-70b-instruct", "name": "Code Llama 70B Instruct", "level": "variant", "family_id": "meta-code-llama", "developer": [ "meta" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 68976910336, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights restricted", "licence": "llama-2-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: code", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Code Llama 70B Instruct, a checkpoint of the Code Llama family. Llama 2 further trained on code, with base, Python and instruction versions from 7B to 70B, for code completion, infilling and chat about code.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2308.12950", "https://huggingface.co/meta-llama/CodeLlama-70b-Instruct-hf", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://www.kaggle.com/models/metaresearch/codellama" ], "unverified": [] }, { "id": "google-codegemma", "name": "CodeGemma", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-03-21", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Gemma models further trained on code: a 2B model for fast completion and infilling, and 7B pre-trained and instruction-tuned models for code generation and chat.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2406.11409", "https://ai.google.dev/gemma/docs/codegemma", "https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/google-models", "https://www.kaggle.com/models/google/codegemma", "https://huggingface.co/google/codegemma-2b", "https://huggingface.co/google/codegemma-7b" ], "unverified": [] }, { "id": "google-codegemma-2b", "name": "CodeGemma 2B", "level": "variant", "family_id": "google-codegemma", "developer": [ "google" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 2506172416, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-03-21", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "CodeGemma 2B, a checkpoint of the CodeGemma family. Gemma models further trained on code: a 2B model for fast completion and infilling, and 7B pre-trained and instruction-tuned models for code generation and chat.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2406.11409", "https://huggingface.co/google/codegemma-2b", "https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/google-models", "https://www.kaggle.com/models/google/codegemma" ], "unverified": [] }, { "id": "google-codegemma-7b", "name": "CodeGemma 7B", "level": "variant", "family_id": "google-codegemma", "developer": [ "google" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8537680896, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-03-21", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "CodeGemma 7B, a checkpoint of the CodeGemma family. Gemma models further trained on code: a 2B model for fast completion and infilling, and 7B pre-trained and instruction-tuned models for code generation and chat.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2406.11409", "https://huggingface.co/google/codegemma-7b", "https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/google-models", "https://www.kaggle.com/models/google/codegemma" ], "unverified": [] }, { "id": "google-codegemma-7b-it", "name": "CodeGemma 7B IT", "level": "variant", "family_id": "google-codegemma", "developer": [ "google" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8537680896, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-03-21", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "CodeGemma 7B IT, a checkpoint of the CodeGemma family. Gemma models further trained on code: a 2B model for fast completion and infilling, and 7B pre-trained and instruction-tuned models for code generation and chat.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2406.11409", "https://huggingface.co/google/codegemma-7b-it", "https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/google-models", "https://www.kaggle.com/models/google/codegemma" ], "unverified": [] }, { "id": "salesforce-codegen", "name": "CodeGen", "level": "family", "family_id": null, "developer": [ "salesforce" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-03-25", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: code", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open code generation models trained on natural language and multiple programming languages, used for program synthesis from prompts.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2203.13474", "https://github.com/salesforce/CodeGen", "https://huggingface.co/OpenVINO/codegen-6B-multi-int4-ov", "https://huggingface.co/Salesforce/codegen-6B-multi", "https://huggingface.co/Salesforce/codegen25-7b-multi_P" ], "unverified": [] }, { "id": "salesforce-codegen-6b-multi", "name": "CodeGen 6B multi", "level": "variant", "family_id": "salesforce-codegen", "developer": [ "salesforce" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-04-13", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "bsd-3-clause", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "CodeGen 6B multi, a checkpoint of the CodeGen family. Open code generation models trained on natural language and multiple programming languages, used for program synthesis from prompts.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2203.13474", "https://huggingface.co/Salesforce/codegen-6B-multi", "https://huggingface.co/OpenVINO/codegen-6B-multi-int4-ov" ], "unverified": [] }, { "id": "salesforce-codegen25-7b-multi", "name": "CodeGen2.5 7B multi", "level": "variant", "family_id": "salesforce-codegen", "developer": [ "salesforce" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-07-06", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: code", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "CodeGen2.5 7B multi, a checkpoint of the CodeGen family. Open code generation models trained on natural language and multiple programming languages, used for program synthesis from prompts.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2203.13474", "https://huggingface.co/Salesforce/codegen25-7b-multi_P", "https://huggingface.co/OpenVINO/codegen-6B-multi-int4-ov" ], "unverified": [] }, { "id": "salesforce-codet5-base", "name": "CodeT5 base", "level": "model", "family_id": null, "developer": [ "salesforce" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-09-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Identifier-aware T5 encoder-decoder pre-trained on code, for code summarisation, generation and translation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2109.00859", "https://huggingface.co/Salesforce/codet5-base", "https://paperswithcode.com/method/codet5" ], "unverified": [] }, { "id": "ibm-granite-code", "name": "Granite Code", "level": "family", "family_id": null, "developer": [ "ibm" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-04-26", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "IBM decoder-only code models from 3B to 34B trained on code in 116 programming languages, for generation, fixing and explaining code.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2405.04324", "https://github.com/ibm-granite/granite-code-models", "https://replicate.com/ibm-granite/granite-8b-code-instruct-128k", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html", "https://huggingface.co/ibm-granite/granite-8b-code-instruct-128k", "https://huggingface.co/ibm-granite/granite-20b-code-instruct-8k" ], "unverified": [] }, { "id": "ibm-granite-8b-code-instruct-128k", "name": "Granite 8B Code Instruct 128K", "level": "variant", "family_id": "ibm-granite-code", "developer": [ "ibm" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8054910976, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-07-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Granite 8B Code Instruct 128K, a checkpoint of the Granite Code family. IBM decoder-only code models from 3B to 34B trained on code in 116 programming languages, for generation, fixing and explaining code.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2405.04324", "https://huggingface.co/ibm-granite/granite-8b-code-instruct-128k", "https://replicate.com/ibm-granite/granite-8b-code-instruct-128k", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html" ], "unverified": [] }, { "id": "ibm-granite-20b-code-instruct-8k", "name": "Granite 20B Code Instruct 8K", "level": "variant", "family_id": "ibm-granite-code", "developer": [ "ibm" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 20066993152, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-04-26", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Granite 20B Code Instruct 8K, a checkpoint of the Granite Code family. IBM decoder-only code models from 3B to 34B trained on code in 116 programming languages, for generation, fixing and explaining code.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2405.04324", "https://huggingface.co/ibm-granite/granite-20b-code-instruct-8k", "https://replicate.com/ibm-granite/granite-8b-code-instruct-128k", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html" ], "unverified": [] }, { "id": "ibm-granite-34b-code-instruct-8k", "name": "Granite 34B Code Instruct 8K", "level": "variant", "family_id": "ibm-granite-code", "developer": [ "ibm" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 33715603456, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-04-26", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Granite 34B Code Instruct 8K, a checkpoint of the Granite Code family. IBM decoder-only code models from 3B to 34B trained on code in 116 programming languages, for generation, fixing and explaining code.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2405.04324", "https://huggingface.co/ibm-granite/granite-34b-code-instruct-8k", "https://replicate.com/ibm-granite/granite-8b-code-instruct-128k", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html" ], "unverified": [] }, { "id": "apple-diffucoder-7b-instruct", "name": "DiffuCoder 7B Instruct", "level": "model", "family_id": null, "developer": [ "apple" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "diffusion", "architecture_detail": null, "params_total": 7615616512, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-07-01", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "apple-amlr", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Masked diffusion language model for code generation that decodes tokens in parallel rather than strictly left to right.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/apple/DiffuCoder-7B-Instruct" ], "unverified": [] }, { "id": "ibm-granite-3", "name": "Granite 3", "level": "family", "family_id": null, "developer": [ "ibm" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-10-02", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "IBM open enterprise language models (2B and 8B dense, small MoE) released under Apache 2.0, updated through versions 3.0 to 3.3; Granite 4 records already in AURION.", "verified_on": "2026-10-12", "sources": [ "https://github.com/ibm-granite/granite-3.0-language-models", "https://replicate.com/ibm-granite/granite-3.3-8b-instruct", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html", "https://huggingface.co/ibm-granite/granite-3.0-8b-instruct", "https://huggingface.co/ibm-granite/granite-3.1-8b-instruct" ], "unverified": [] }, { "id": "ibm-granite-3-0-8b-instruct", "name": "Granite 3.0 8B Instruct", "level": "variant", "family_id": "ibm-granite-3", "developer": [ "ibm" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8170848256, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-10-02", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Granite 3.0 8B Instruct, a checkpoint of the Granite 3 family. IBM open enterprise language models (2B and 8B dense, small MoE) released under Apache 2.0, updated through versions 3.0 to 3.3; Granite 4 records already in AURION.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/ibm-granite/granite-3.0-8b-instruct", "https://replicate.com/ibm-granite/granite-3.3-8b-instruct", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html" ], "unverified": [] }, { "id": "ibm-granite-3-1-8b-instruct", "name": "Granite 3.1 8B Instruct", "level": "variant", "family_id": "ibm-granite-3", "developer": [ "ibm" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8170848256, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-12-06", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Granite 3.1 8B Instruct, a checkpoint of the Granite 3 family. IBM open enterprise language models (2B and 8B dense, small MoE) released under Apache 2.0, updated through versions 3.0 to 3.3; Granite 4 records already in AURION.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/ibm-granite/granite-3.1-8b-instruct", "https://replicate.com/ibm-granite/granite-3.3-8b-instruct", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html" ], "unverified": [] }, { "id": "ibm-granite-3-2-8b-instruct", "name": "Granite 3.2 8B Instruct", "level": "variant", "family_id": "ibm-granite-3", "developer": [ "ibm" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8170848256, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-02-17", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Granite 3.2 8B Instruct, a checkpoint of the Granite 3 family. IBM open enterprise language models (2B and 8B dense, small MoE) released under Apache 2.0, updated through versions 3.0 to 3.3; Granite 4 records already in AURION.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/ibm-granite/granite-3.2-8b-instruct", "https://replicate.com/ibm-granite/granite-3.3-8b-instruct", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html" ], "unverified": [] }, { "id": "ibm-granite-3-3-8b-instruct", "name": "Granite 3.3 8B Instruct", "level": "variant", "family_id": "ibm-granite-3", "developer": [ "ibm" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8170864640, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-04-09", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Granite 3.3 8B Instruct, a checkpoint of the Granite 3 family. IBM open enterprise language models (2B and 8B dense, small MoE) released under Apache 2.0, updated through versions 3.0 to 3.3; Granite 4 records already in AURION.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/ibm-granite/granite-3.3-8b-instruct", "https://replicate.com/ibm-granite/granite-3.3-8b-instruct", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html" ], "unverified": [] }, { "id": "ibm-granite-embedding", "name": "Granite Embedding", "level": "family", "family_id": null, "developer": [ "ibm" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "The multilingual R2 embedder is a separate AURION record (ibm-granite-embedding-multilingual-r2).", "context_length": null, "release_date": "2024-12-04", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, ar, cs, de, es, fr, it, ja, ko, nl, pt, zh", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "IBM open retrieval models for enterprise search: English and multilingual embedders and a reranker, trained on permissively licensed data.", "verified_on": "2026-10-12", "sources": [ "https://replicate.com/ibm-granite/granite-embedding-278m-multilingual", "https://ai.azure.com/catalog/models?task=rerank", "https://huggingface.co/ibm-granite/granite-embedding-278m-multilingual", "https://huggingface.co/ibm-granite/granite-embedding-small-english-r2" ], "unverified": [] }, { "id": "ibm-granite-embedding-278m-multilingual", "name": "Granite Embedding 278M multilingual", "level": "variant", "family_id": "ibm-granite-embedding", "developer": [ "ibm" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 278043648, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-12-04", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, ar, cs, de, es, fr, it, ja, ko, nl, pt, zh", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Granite Embedding 278M multilingual, a checkpoint of the Granite Embedding family. IBM open retrieval models for enterprise search: English and multilingual embedders and a reranker, trained on permissively licensed data.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/ibm-granite/granite-embedding-278m-multilingual", "https://replicate.com/ibm-granite/granite-embedding-278m-multilingual", "https://ai.azure.com/catalog/models?task=rerank" ], "unverified": [] }, { "id": "ibm-granite-embedding-small-english-r2", "name": "Granite Embedding small English R2", "level": "variant", "family_id": "ibm-granite-embedding", "developer": [ "ibm" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 47662464, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-07-17", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Granite Embedding small English R2, a checkpoint of the Granite Embedding family. IBM open retrieval models for enterprise search: English and multilingual embedders and a reranker, trained on permissively licensed data.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/ibm-granite/granite-embedding-small-english-r2", "https://replicate.com/ibm-granite/granite-embedding-278m-multilingual", "https://ai.azure.com/catalog/models?task=rerank" ], "unverified": [] }, { "id": "ibm-granite-embedding-reranker-english-r2", "name": "Granite Embedding Reranker English R2", "level": "variant", "family_id": "ibm-granite-embedding", "developer": [ "ibm" ], "model_class": "reranker", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 149605633, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-08-04", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Granite Embedding Reranker English R2, a checkpoint of the Granite Embedding family. IBM open retrieval models for enterprise search: English and multilingual embedders and a reranker, trained on permissively licensed data.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/ibm-granite/granite-embedding-reranker-english-r2", "https://replicate.com/ibm-granite/granite-embedding-278m-multilingual", "https://ai.azure.com/catalog/models?task=rerank" ], "unverified": [] }, { "id": "google-recurrentgemma", "name": "RecurrentGemma", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Griffin: linear recurrences mixed with local attention", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-04-06", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open models on the Griffin architecture, which mixes linear recurrence with local attention so memory stays fixed on long sequences.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2404.07839", "https://www.kaggle.com/models/google/recurrentgemma", "https://huggingface.co/google/recurrentgemma-2b", "https://huggingface.co/google/recurrentgemma-9b" ], "unverified": [] }, { "id": "google-recurrentgemma-2b", "name": "RecurrentGemma 2B", "level": "variant", "family_id": "google-recurrentgemma", "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": null, "params_total": 2682862080, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-04-06", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "RecurrentGemma 2B, a checkpoint of the RecurrentGemma family. Open models on the Griffin architecture, which mixes linear recurrence with local attention so memory stays fixed on long sequences.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2404.07839", "https://huggingface.co/google/recurrentgemma-2b", "https://www.kaggle.com/models/google/recurrentgemma" ], "unverified": [] }, { "id": "google-recurrentgemma-9b", "name": "RecurrentGemma 9B", "level": "variant", "family_id": "google-recurrentgemma", "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": null, "params_total": 9628553216, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-06-07", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "RecurrentGemma 9B, a checkpoint of the RecurrentGemma family. Open models on the Griffin architecture, which mixes linear recurrence with local attention so memory stays fixed on long sequences.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2404.07839", "https://huggingface.co/google/recurrentgemma-9b", "https://www.kaggle.com/models/google/recurrentgemma" ], "unverified": [] }, { "id": "google-paligemma", "name": "PaliGemma", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-05-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open 3B vision language model combining a SigLIP image encoder with Gemma, built to be fine-tuned for captioning, visual question answering, detection and OCR.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2407.07726", "https://ai.google.dev/gemma/docs/paligemma", "https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/google-models", "https://huggingface.co/api/models?pipeline_tag=visual-document-retrieval&sort=downloads&limit=50", "https://huggingface.co/google/paligemma-3b-pt-224", "https://huggingface.co/google/paligemma-3b-mix-448" ], "unverified": [] }, { "id": "google-paligemma-3b-pt-224", "name": "PaliGemma 3B pt 224", "level": "variant", "family_id": "google-paligemma", "developer": [ "google" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 2923466480, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-05-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "PaliGemma 3B pt 224, a checkpoint of the PaliGemma family. Open 3B vision language model combining a SigLIP image encoder with Gemma, built to be fine-tuned for captioning, visual question answering, detection and OCR.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2407.07726", "https://huggingface.co/google/paligemma-3b-pt-224", "https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/google-models", "https://huggingface.co/api/models?pipeline_tag=visual-document-retrieval&sort=downloads&limit=50" ], "unverified": [] }, { "id": "google-paligemma-3b-mix-448", "name": "PaliGemma 3B mix 448", "level": "variant", "family_id": "google-paligemma", "developer": [ "google" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 2924351216, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-05-13", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "PaliGemma 3B mix 448, a checkpoint of the PaliGemma family. Open 3B vision language model combining a SigLIP image encoder with Gemma, built to be fine-tuned for captioning, visual question answering, detection and OCR.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2407.07726", "https://huggingface.co/google/paligemma-3b-mix-448", "https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/google-models", "https://huggingface.co/api/models?pipeline_tag=visual-document-retrieval&sort=downloads&limit=50" ], "unverified": [] }, { "id": "google-paligemma-2", "name": "PaliGemma 2", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-11-21", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "PaliGemma upgraded to Gemma 2 language models in 3B, 10B and 28B sizes at three image resolutions, for transfer to OCR, captioning and medical imaging tasks.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2412.03555", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://docs.roboflow.com/models/supported-models/paligemma2.md", "https://huggingface.co/google/paligemma2-3b-pt-224", "https://huggingface.co/google/paligemma2-10b-pt-448" ], "unverified": [] }, { "id": "google-paligemma2-3b-pt-224", "name": "PaliGemma 2 3B pt 224", "level": "variant", "family_id": "google-paligemma-2", "developer": [ "google" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 3032242416, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-11-21", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "PaliGemma 2 3B pt 224, a checkpoint of the PaliGemma 2 family. PaliGemma upgraded to Gemma 2 language models in 3B, 10B and 28B sizes at three image resolutions, for transfer to OCR, captioning and medical imaging tasks.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2412.03555", "https://huggingface.co/google/paligemma2-3b-pt-224", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://docs.roboflow.com/models/supported-models/paligemma2.md" ], "unverified": [] }, { "id": "google-paligemma2-10b-pt-448", "name": "PaliGemma 2 10B pt 448", "level": "variant", "family_id": "google-paligemma-2", "developer": [ "google" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 9663523568, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-11-21", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "PaliGemma 2 10B pt 448, a checkpoint of the PaliGemma 2 family. PaliGemma upgraded to Gemma 2 language models in 3B, 10B and 28B sizes at three image resolutions, for transfer to OCR, captioning and medical imaging tasks.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2412.03555", "https://huggingface.co/google/paligemma2-10b-pt-448", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://docs.roboflow.com/models/supported-models/paligemma2.md" ], "unverified": [] }, { "id": "google-paligemma2-28b-pt-896", "name": "PaliGemma 2 28B pt 896", "level": "variant", "family_id": "google-paligemma-2", "developer": [ "google" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 27654910704, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-11-22", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "PaliGemma 2 28B pt 896, a checkpoint of the PaliGemma 2 family. PaliGemma upgraded to Gemma 2 language models in 3B, 10B and 28B sizes at three image resolutions, for transfer to OCR, captioning and medical imaging tasks.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2412.03555", "https://huggingface.co/google/paligemma2-28b-pt-896", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://docs.roboflow.com/models/supported-models/paligemma2.md" ], "unverified": [] }, { "id": "google-datagemma", "name": "DataGemma", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-08-26", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Gemma 2 27B models fine-tuned to ground answers in public statistics from Google Data Commons, by interleaving queries (RIG) or by retrieval (RAG).", "verified_on": "2026-10-12", "sources": [ "https://www.kaggle.com/models/google/datagemma-rig", "https://www.kaggle.com/models/google/datagemma-rag", "https://huggingface.co/google/datagemma-rig-27b-it", "https://huggingface.co/google/datagemma-rag-27b-it" ], "unverified": [] }, { "id": "google-datagemma-rig-27b-it", "name": "DataGemma RIG 27B IT", "level": "variant", "family_id": "google-datagemma", "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 27227128320, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-08-27", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DataGemma RIG 27B IT, a checkpoint of the DataGemma family. Gemma 2 27B models fine-tuned to ground answers in public statistics from Google Data Commons, by interleaving queries (RIG) or by retrieval (RAG).", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/google/datagemma-rig-27b-it", "https://www.kaggle.com/models/google/datagemma-rig", "https://www.kaggle.com/models/google/datagemma-rag" ], "unverified": [] }, { "id": "google-datagemma-rag-27b-it", "name": "DataGemma RAG 27B IT", "level": "variant", "family_id": "google-datagemma", "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 27227128320, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-08-26", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DataGemma RAG 27B IT, a checkpoint of the DataGemma family. Gemma 2 27B models fine-tuned to ground answers in public statistics from Google Data Commons, by interleaving queries (RIG) or by retrieval (RAG).", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/google/datagemma-rag-27b-it", "https://www.kaggle.com/models/google/datagemma-rig", "https://www.kaggle.com/models/google/datagemma-rag" ], "unverified": [] }, { "id": "google-vaultgemma-1b", "name": "VaultGemma 1B", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 1038741120, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-09-05", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "1-billion-parameter Gemma model trained from scratch with differential privacy, so it gives formal privacy guarantees on its training data.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/google/vaultgemma-1b", "https://www.kaggle.com/models/google/vaultgemma", "https://www.kaggle.com/models/keras/vaultgemma" ], "unverified": [] }, { "id": "google-t5gemma", "name": "T5Gemma", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-04-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Encoder-decoder models made by adapting pre-trained decoder-only Gemma 2 checkpoints, for summarisation, translation and other input-heavy tasks.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2504.06225", "https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/google-models", "https://keras.io/keras_hub/presets/", "https://huggingface.co/google/t5gemma-2b-2b-ul2", "https://huggingface.co/google/t5gemma-9b-9b-ul2" ], "unverified": [] }, { "id": "google-t5gemma-2b-2b-ul2", "name": "T5Gemma 2B-2B UL2", "level": "variant", "family_id": "google-t5gemma", "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 5596853760, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-06-19", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "T5Gemma 2B-2B UL2, a checkpoint of the T5Gemma family. Encoder-decoder models made by adapting pre-trained decoder-only Gemma 2 checkpoints, for summarisation, translation and other input-heavy tasks.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2504.06225", "https://huggingface.co/google/t5gemma-2b-2b-ul2", "https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/google-models", "https://keras.io/keras_hub/presets/" ], "unverified": [] }, { "id": "google-t5gemma-9b-9b-ul2", "name": "T5Gemma 9B-9B UL2", "level": "variant", "family_id": "google-t5gemma", "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 20333401088, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-06-19", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "T5Gemma 9B-9B UL2, a checkpoint of the T5Gemma family. Encoder-decoder models made by adapting pre-trained decoder-only Gemma 2 checkpoints, for summarisation, translation and other input-heavy tasks.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2504.06225", "https://huggingface.co/google/t5gemma-9b-9b-ul2", "https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/google-models", "https://keras.io/keras_hub/presets/" ], "unverified": [] }, { "id": "google-t5gemma-2", "name": "T5Gemma 2", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10-25", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Second-generation encoder-decoder models adapted from Gemma 3, adding image input, long context and multilingual support.", "verified_on": "2026-10-12", "sources": [ "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://huggingface.co/api/models?pipeline_tag=zero-shot-classification&sort=likes&limit=50", "https://huggingface.co/google/t5gemma-2-270m-270m", "https://huggingface.co/google/t5gemma-2-4b-4b" ], "unverified": [] }, { "id": "google-t5gemma-2-270m-270m", "name": "T5Gemma 2 270M-270M", "level": "variant", "family_id": "google-t5gemma-2", "developer": [ "google" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 786029296, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-10-25", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "T5Gemma 2 270M-270M, a checkpoint of the T5Gemma 2 family. Second-generation encoder-decoder models adapted from Gemma 3, adding image input, long context and multilingual support.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/google/t5gemma-2-270m-270m", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://huggingface.co/api/models?pipeline_tag=zero-shot-classification&sort=likes&limit=50" ], "unverified": [] }, { "id": "google-t5gemma-2-4b-4b", "name": "T5Gemma 2 4B-4B", "level": "variant", "family_id": "google-t5gemma-2", "developer": [ "google" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8851108720, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-10-25", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "T5Gemma 2 4B-4B, a checkpoint of the T5Gemma 2 family. Second-generation encoder-decoder models adapted from Gemma 3, adding image input, long context and multilingual support.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/google/t5gemma-2-4b-4b", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://huggingface.co/api/models?pipeline_tag=zero-shot-classification&sort=likes&limit=50" ], "unverified": [] }, { "id": "google-gemma-2-2b-jpn-it", "name": "Gemma 2 2B JPN IT", "level": "model", "family_id": "google-gemma-2", "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 2614341888, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-09-25", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Japanese and English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Gemma 2 2B fine-tuned on Japanese text so it performs in Japanese at the level of the English model.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/google/gemma-2-2b-jpn-it", "https://www.kaggle.com/models/google/gemma-2-2b-jpn-it" ], "unverified": [] }, { "id": "google-translategemma", "name": "TranslateGemma", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "translation", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open translation models built on Gemma 3 that translate text, and text in images, across 55 languages.", "verified_on": "2026-10-12", "sources": [ "https://www.kaggle.com/models/google/translategemma", "https://www.kaggle.com/models/keras/translategemma", "https://huggingface.co/google/translategemma-4b-it", "https://huggingface.co/google/translategemma-12b-it" ], "unverified": [] }, { "id": "google-translategemma-4b-it", "name": "TranslateGemma 4B", "level": "variant", "family_id": "google-translategemma", "developer": [ "google" ], "model_class": "translation", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 4971331952, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2026-01-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "TranslateGemma 4B, a checkpoint of the TranslateGemma family. Open translation models built on Gemma 3 that translate text, and text in images, across 55 languages.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/google/translategemma-4b-it", "https://www.kaggle.com/models/google/translategemma", "https://www.kaggle.com/models/keras/translategemma" ], "unverified": [] }, { "id": "google-translategemma-12b-it", "name": "TranslateGemma 12B", "level": "variant", "family_id": "google-translategemma", "developer": [ "google" ], "model_class": "translation", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 13194203760, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2026-01-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "TranslateGemma 12B, a checkpoint of the TranslateGemma family. Open translation models built on Gemma 3 that translate text, and text in images, across 55 languages.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/google/translategemma-12b-it", "https://www.kaggle.com/models/google/translategemma", "https://www.kaggle.com/models/keras/translategemma" ], "unverified": [] }, { "id": "google-translategemma-27b-it", "name": "TranslateGemma 27B", "level": "variant", "family_id": "google-translategemma", "developer": [ "google" ], "model_class": "translation", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 28842036848, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2026-01-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "gemma-terms", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "TranslateGemma 27B, a checkpoint of the TranslateGemma family. Open translation models built on Gemma 3 that translate text, and text in images, across 55 languages.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/google/translategemma-27b-it", "https://www.kaggle.com/models/google/translategemma", "https://www.kaggle.com/models/keras/translategemma" ], "unverified": [] }, { "id": "google-diffusiongemma-26b-a4b-it", "name": "DiffusionGemma 26B A4B", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "diffusion", "architecture_detail": null, "params_total": 25823778864, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2026-06-09", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open text diffusion language model from the Gemma line that generates by iteratively denoising blocks of tokens.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/google/diffusiongemma-26B-A4B-it", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://www.kaggle.com/models/google/diffusiongemma" ], "unverified": [] }, { "id": "meta-nllb-200", "name": "NLLB-200", "level": "family", "family_id": null, "developer": [ "meta" ], "model_class": "translation", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Dense 600M (distilled), 1.3B (and distilled), 3.3B, and the 54.5B MoE research model.", "context_length": null, "release_date": "2022-07-08", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Checkpoint model cards list 196 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Acehnese", "Balinese", "Banjar", "Buginese", "Javanese", "Khmer", "Lao", "Minangkabau", "Burmese", "Sundanese", "Tagalog", "Thai", "Vietnamese", "Malay" ], "indonesia_hosting": "unknown", "status": "active", "summary": "No Language Left Behind: one translation model covering 200 languages, many of them low-resource, released for research under a non-commercial licence.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2207.04672", "https://github.com/facebookresearch/fairseq/tree/nllb", "https://huggingface.co/api/models?pipeline_tag=translation&sort=downloads&limit=50", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://huggingface.co/facebook/nllb-200-distilled-600M", "https://huggingface.co/facebook/nllb-200-distilled-1.3B" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "meta-nllb-200-distilled-600m", "name": "NLLB-200 distilled 600M", "level": "variant", "family_id": "meta-nllb-200", "developer": [ "meta" ], "model_class": "translation", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-07-08", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 196 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Acehnese", "Balinese", "Banjar", "Buginese", "Javanese", "Khmer", "Lao", "Minangkabau", "Burmese", "Sundanese", "Tagalog", "Thai", "Vietnamese", "Malay" ], "indonesia_hosting": "unknown", "status": "active", "summary": "NLLB-200 distilled 600M, a checkpoint of the NLLB-200 family. No Language Left Behind: one translation model covering 200 languages, many of them low-resource, released for research under a non-commercial licence.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2207.04672", "https://huggingface.co/facebook/nllb-200-distilled-600M", "https://huggingface.co/api/models?pipeline_tag=translation&sort=downloads&limit=50", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "meta-nllb-200-distilled-1-3b", "name": "NLLB-200 distilled 1.3B", "level": "variant", "family_id": "meta-nllb-200", "developer": [ "meta" ], "model_class": "translation", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-07-08", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 196 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Acehnese", "Balinese", "Banjar", "Buginese", "Javanese", "Khmer", "Lao", "Minangkabau", "Burmese", "Sundanese", "Tagalog", "Thai", "Vietnamese", "Malay" ], "indonesia_hosting": "unknown", "status": "active", "summary": "NLLB-200 distilled 1.3B, a checkpoint of the NLLB-200 family. No Language Left Behind: one translation model covering 200 languages, many of them low-resource, released for research under a non-commercial licence.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2207.04672", "https://huggingface.co/facebook/nllb-200-distilled-1.3B", "https://huggingface.co/api/models?pipeline_tag=translation&sort=downloads&limit=50", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "meta-nllb-200-3-3b", "name": "NLLB-200 3.3B", "level": "variant", "family_id": "meta-nllb-200", "developer": [ "meta" ], "model_class": "translation", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-07-08", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 196 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Acehnese", "Balinese", "Banjar", "Buginese", "Javanese", "Khmer", "Lao", "Minangkabau", "Burmese", "Sundanese", "Tagalog", "Thai", "Vietnamese", "Malay" ], "indonesia_hosting": "unknown", "status": "active", "summary": "NLLB-200 3.3B, a checkpoint of the NLLB-200 family. No Language Left Behind: one translation model covering 200 languages, many of them low-resource, released for research under a non-commercial licence.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2207.04672", "https://huggingface.co/facebook/nllb-200-3.3B", "https://huggingface.co/api/models?pipeline_tag=translation&sort=downloads&limit=50", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "university-of-helsinki-opus-mt", "name": "OPUS-MT", "level": "family", "family_id": null, "developer": [ "university-of-helsinki" ], "model_class": "translation", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server", "laptop" ], "footprint_note": null, "runtimes": [], "languages_note": "Many language pairs, including English-Indonesian and Indonesian-English", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small Marian translation models, one per language pair, trained on the open OPUS parallel corpora and runnable on CPU.", "verified_on": "2026-10-12", "sources": [ "https://github.com/Helsinki-NLP/Opus-MT", "https://huggingface.co/api/models?pipeline_tag=translation&sort=downloads&limit=50", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://huggingface.co/Helsinki-NLP/opus-mt-en-id", "https://huggingface.co/Helsinki-NLP/opus-mt-id-en" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "university-of-helsinki-opus-mt-en-id", "name": "OPUS-MT en-id", "level": "variant", "family_id": "university-of-helsinki-opus-mt", "developer": [ "university-of-helsinki" ], "model_class": "translation", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English to Indonesian", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Marian translation model from English to Indonesian, trained on OPUS data; one of the OPUS-MT language-pair models.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Helsinki-NLP/opus-mt-en-id", "https://huggingface.co/api/models?pipeline_tag=translation&sort=downloads&limit=50", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "university-of-helsinki-opus-mt-id-en", "name": "OPUS-MT id-en", "level": "variant", "family_id": "university-of-helsinki-opus-mt", "developer": [ "university-of-helsinki" ], "model_class": "translation", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Indonesian to English", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Marian translation model from Indonesian to English, trained on OPUS data; one of the OPUS-MT language-pair models.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Helsinki-NLP/opus-mt-id-en", "https://huggingface.co/api/models?pipeline_tag=translation&sort=downloads&limit=50", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "university-of-helsinki-opus-mt-en-zh", "name": "OPUS-MT en-zh", "level": "variant", "family_id": "university-of-helsinki-opus-mt", "developer": [ "university-of-helsinki" ], "model_class": "translation", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, zh", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OPUS-MT en-zh, a checkpoint of the OPUS-MT family. Small Marian translation models, one per language pair, trained on the open OPUS parallel corpora and runnable on CPU.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Helsinki-NLP/opus-mt-en-zh", "https://huggingface.co/api/models?pipeline_tag=translation&sort=downloads&limit=50", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "google-madlad-400", "name": "MADLAD-400 MT", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "translation", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-09-09", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Checkpoint model cards list 418 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Vietnamese", "Thai", "Malay", "Burmese", "Khmer", "Lao", "Javanese", "Sundanese", "Minangkabau", "Acehnese", "Balinese", "Madurese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "T5-based translation models trained on the MADLAD-400 web corpus, translating across more than 400 languages.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2309.04662", "https://huggingface.co/api/models?pipeline_tag=translation&sort=likes&limit=50", "https://ai.azure.com/catalog/models?task=object-detection", "https://huggingface.co/google/madlad400-3b-mt", "https://huggingface.co/google/madlad400-10b-mt" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "google-madlad400-3b-mt", "name": "MADLAD-400 3B MT", "level": "variant", "family_id": "google-madlad-400", "developer": [ "google" ], "model_class": "translation", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 2940374016, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-11-27", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 419 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Vietnamese", "Thai", "Malay", "Burmese", "Khmer", "Lao", "Javanese", "Sundanese", "Minangkabau", "Acehnese", "Balinese", "Madurese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "MADLAD-400 3B MT, a checkpoint of the MADLAD-400 MT family. T5-based translation models trained on the MADLAD-400 web corpus, translating across more than 400 languages.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2309.04662", "https://huggingface.co/google/madlad400-3b-mt", "https://huggingface.co/api/models?pipeline_tag=translation&sort=likes&limit=50", "https://ai.azure.com/catalog/models?task=object-detection" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "google-madlad400-10b-mt", "name": "MADLAD-400 10B MT", "level": "variant", "family_id": "google-madlad-400", "developer": [ "google" ], "model_class": "translation", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 10712586240, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-11-27", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 419 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Vietnamese", "Thai", "Malay", "Burmese", "Khmer", "Lao", "Javanese", "Sundanese", "Minangkabau", "Acehnese", "Balinese", "Madurese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "MADLAD-400 10B MT, a checkpoint of the MADLAD-400 MT family. T5-based translation models trained on the MADLAD-400 web corpus, translating across more than 400 languages.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2309.04662", "https://huggingface.co/google/madlad400-10b-mt", "https://huggingface.co/api/models?pipeline_tag=translation&sort=likes&limit=50", "https://ai.azure.com/catalog/models?task=object-detection" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "ai4bharat-indictrans2-en-indic-1b", "name": "IndicTrans2 (en-indic 1B)", "level": "model", "family_id": null, "developer": [ "ai4bharat" ], "model_class": "translation", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 1115543552, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-09-09", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "22 scheduled Indian languages and English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Translation models between English and all 22 scheduled Indian languages; distilled 200M versions are also released.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/ai4bharat/indictrans2-en-indic-1B", "https://huggingface.co/api/models?pipeline_tag=translation&sort=downloads&limit=50" ], "unverified": [] }, { "id": "tencent-hunyuan-mt", "name": "Hunyuan-MT", "level": "family", "family_id": null, "developer": [ "tencent" ], "model_class": "translation", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-08-28", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Checkpoint model cards list 36 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Thai", "Vietnamese", "Malay", "Tagalog", "Khmer", "Burmese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Tencent translation language models covering 33 languages, including several Chinese minority languages; Chimera fuses several candidate translations into one.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/api/models?pipeline_tag=translation&sort=likes&limit=50", "https://www.modelscope.cn/models/Tencent-Hunyuan/Hunyuan-MT-7B", "https://huggingface.co/tencent/Hunyuan-MT-7B", "https://huggingface.co/tencent/Hunyuan-MT-Chimera-7B" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "tencent-hunyuan-mt-7b", "name": "Hunyuan-MT-7B", "level": "variant", "family_id": "tencent-hunyuan-mt", "developer": [ "tencent" ], "model_class": "translation", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8030269440, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-08-28", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Hunyuan-MT-7B, a checkpoint of the Hunyuan-MT family. Tencent translation language models covering 33 languages, including several Chinese minority languages; Chimera fuses several candidate translations into one.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/tencent/Hunyuan-MT-7B", "https://huggingface.co/api/models?pipeline_tag=translation&sort=likes&limit=50", "https://www.modelscope.cn/models/Tencent-Hunyuan/Hunyuan-MT-7B" ], "unverified": [] }, { "id": "tencent-hunyuan-mt-chimera-7b", "name": "Hunyuan-MT-Chimera-7B", "level": "variant", "family_id": "tencent-hunyuan-mt", "developer": [ "tencent" ], "model_class": "translation", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8030269440, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-08-28", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 36 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Thai", "Vietnamese", "Malay", "Tagalog", "Khmer", "Burmese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Hunyuan-MT-Chimera-7B, a checkpoint of the Hunyuan-MT family. Tencent translation language models covering 33 languages, including several Chinese minority languages; Chimera fuses several candidate translations into one.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/tencent/Hunyuan-MT-Chimera-7B", "https://huggingface.co/api/models?pipeline_tag=translation&sort=likes&limit=50", "https://www.modelscope.cn/models/Tencent-Hunyuan/Hunyuan-MT-7B" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "tencent-hy-mt1-5-1-8b", "name": "HY-MT1.5 1.8B", "level": "variant", "family_id": "tencent-hunyuan-mt", "developer": [ "tencent" ], "model_class": "translation", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 2038515712, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-12-25", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 36 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Thai", "Vietnamese", "Malay", "Tagalog", "Khmer", "Burmese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "HY-MT1.5 1.8B, a checkpoint of the Hunyuan-MT family. Tencent translation language models covering 33 languages, including several Chinese minority languages; Chimera fuses several candidate translations into one.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/tencent/HY-MT1.5-1.8B", "https://huggingface.co/api/models?pipeline_tag=translation&sort=likes&limit=50", "https://www.modelscope.cn/models/Tencent-Hunyuan/Hunyuan-MT-7B" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "tencent-hy-mt2-1-8b", "name": "Hy-MT2 1.8B", "level": "variant", "family_id": "tencent-hunyuan-mt", "developer": [ "tencent" ], "model_class": "translation", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 2038515712, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2026-05-11", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 36 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Thai", "Vietnamese", "Malay", "Tagalog", "Khmer", "Burmese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Hy-MT2 1.8B, a checkpoint of the Hunyuan-MT family. Tencent translation language models covering 33 languages, including several Chinese minority languages; Chimera fuses several candidate translations into one.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/tencent/Hy-MT2-1.8B", "https://huggingface.co/api/models?pipeline_tag=translation&sort=likes&limit=50", "https://www.modelscope.cn/models/Tencent-Hunyuan/Hunyuan-MT-7B" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "bytedance-seed-x-ppo-7b", "name": "Seed-X PPO 7B", "level": "model", "family_id": null, "developer": [ "bytedance" ], "model_class": "translation", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 7514271744, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-07-16", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 28 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Malay", "Thai", "Vietnamese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "7-billion-parameter multilingual translation model from ByteDance Seed, trained with reinforcement learning, covering 28 languages.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/ByteDance-Seed/Seed-X-PPO-7B", "https://huggingface.co/api/models?pipeline_tag=translation&sort=likes&limit=50" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "unbabel-towerinstruct-7b-v0-2", "name": "TowerInstruct 7B v0.2", "level": "model", "family_id": null, "developer": [ "unbabel" ], "model_class": "translation", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 6738472960, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-02-09", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, de, fr, zh, pt, nl, ru, ko, it, es", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Llama 2 based model tuned for translation and related tasks (post-editing, named entities, grammar correction) in 10 languages.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Unbabel/TowerInstruct-7B-v0.2", "https://ai.azure.com/catalog/models?task=translation" ], "unverified": [] }, { "id": "google-transformer-nmt", "name": "Transformer (WMT translation, base and big)", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "translation", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-06-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "The original attention-only encoder-decoder, trained on WMT English-German and English-French translation; the basis of later language models.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1706.03762", "https://github.com/tensorflow/models/tree/master/official/nlp", "https://github.com/tensorflow/models/blob/master/official/README.md", "https://pytorch.org/hub/pytorch_fairseq_translation" ], "unverified": [] }, { "id": "meta-seamless-m4t", "name": "SeamlessM4T", "level": "family", "family_id": null, "developer": [ "meta" ], "model_class": "translation", "modalities_in": [ "text", "audio" ], "modalities_out": [ "text", "audio" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-08-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Checkpoint model cards list 96 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Javanese", "Khmer", "Lao", "Burmese", "Tagalog", "Thai", "Vietnamese", "Malay" ], "indonesia_hosting": "unknown", "status": "active", "summary": "One model for speech-to-speech, speech-to-text, text-to-speech and text-to-text translation and speech recognition for up to 100 languages.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2308.11596", "https://github.com/facebookresearch/seamless_communication", "https://replicate.com/cjwbw/seamless_communication", "https://huggingface.co/facebook/seamless-m4t-v2-large" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "meta-seamless-m4t-v2-large", "name": "SeamlessM4T v2 large", "level": "variant", "family_id": "meta-seamless-m4t", "developer": [ "meta" ], "model_class": "translation", "modalities_in": [ "text", "audio" ], "modalities_out": [ "text", "audio" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 2309249669, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-11-30", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 96 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Javanese", "Khmer", "Lao", "Burmese", "Tagalog", "Thai", "Vietnamese", "Malay" ], "indonesia_hosting": "unknown", "status": "active", "summary": "SeamlessM4T v2 large, a checkpoint of the SeamlessM4T family. One model for speech-to-speech, speech-to-text, text-to-speech and text-to-text translation and speech recognition for up to 100 languages.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2312.05187", "https://huggingface.co/facebook/seamless-m4t-v2-large", "https://replicate.com/cjwbw/seamless_communication", "https://ai.meta.com/blog/seamless-communication/" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "openai-clip", "name": "CLIP", "level": "family", "family_id": null, "developer": [ "openai" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-02-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Image and text encoders trained together on 400 million image-caption pairs, used for zero-shot image classification and image-text search.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.00020", "https://github.com/openai/CLIP", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://huggingface.co/api/models?pipeline_tag=zero-shot-image-classification&sort=downloads&limit=50", "https://huggingface.co/openai/clip-vit-base-patch32", "https://huggingface.co/openai/clip-vit-large-patch14" ], "unverified": [] }, { "id": "openai-clip-vit-base-patch32", "name": "CLIP ViT-B/32", "level": "variant", "family_id": "openai-clip", "developer": [ "openai" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "CLIP ViT-B/32, a checkpoint of the CLIP family. Image and text encoders trained together on 400 million image-caption pairs, used for zero-shot image classification and image-text search.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.00020", "https://huggingface.co/openai/clip-vit-base-patch32", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://huggingface.co/api/models?pipeline_tag=zero-shot-image-classification&sort=downloads&limit=50" ], "unverified": [] }, { "id": "openai-clip-vit-large-patch14", "name": "CLIP ViT-L/14", "level": "variant", "family_id": "openai-clip", "developer": [ "openai" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 427616846, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "CLIP ViT-L/14, a checkpoint of the CLIP family. Image and text encoders trained together on 400 million image-caption pairs, used for zero-shot image classification and image-text search.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.00020", "https://huggingface.co/openai/clip-vit-large-patch14", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://huggingface.co/api/models?pipeline_tag=zero-shot-image-classification&sort=downloads&limit=50" ], "unverified": [] }, { "id": "openai-clip-vit-large-patch14-336", "name": "CLIP ViT-L/14 336px", "level": "variant", "family_id": "openai-clip", "developer": [ "openai" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-04-22", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "CLIP ViT-L/14 336px, a checkpoint of the CLIP family. Image and text encoders trained together on 400 million image-caption pairs, used for zero-shot image classification and image-text search.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.00020", "https://huggingface.co/openai/clip-vit-large-patch14-336", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://huggingface.co/api/models?pipeline_tag=zero-shot-image-classification&sort=downloads&limit=50" ], "unverified": [] }, { "id": "laion-openclip", "name": "OpenCLIP (LAION)", "level": "family", "family_id": null, "developer": [ "laion" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-09-14", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open reproduction of CLIP trained on the public LAION-2B image-text dataset, used for zero-shot classification and retrieval and as the text encoder of image generators.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2212.07143", "https://github.com/mlfoundations/open_clip", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://modelzoo.co/model/open-clip", "https://huggingface.co/laion/CLIP-ViT-B-32-laion2B-s34B-b79K", "https://huggingface.co/laion/CLIP-ViT-H-14-laion2B-s32B-b79K" ], "unverified": [] }, { "id": "laion-clip-vit-b-32-laion2b-s34b-b79k", "name": "OpenCLIP ViT-B-32 LAION-2B", "level": "variant", "family_id": "laion-openclip", "developer": [ "laion" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 151277439, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2022-09-14", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenCLIP ViT-B-32 LAION-2B, a checkpoint of the OpenCLIP (LAION) family. Open reproduction of CLIP trained on the public LAION-2B image-text dataset, used for zero-shot classification and retrieval and as the text encoder of image generators.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2212.07143", "https://huggingface.co/laion/CLIP-ViT-B-32-laion2B-s34B-b79K", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://modelzoo.co/model/open-clip" ], "unverified": [] }, { "id": "laion-clip-vit-h-14-laion2b-s32b-b79k", "name": "OpenCLIP ViT-H-14 LAION-2B", "level": "variant", "family_id": "laion-openclip", "developer": [ "laion" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 986109774, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2022-09-14", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenCLIP ViT-H-14 LAION-2B, a checkpoint of the OpenCLIP (LAION) family. Open reproduction of CLIP trained on the public LAION-2B image-text dataset, used for zero-shot classification and retrieval and as the text encoder of image generators.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2212.07143", "https://huggingface.co/laion/CLIP-ViT-H-14-laion2B-s32B-b79K", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://modelzoo.co/model/open-clip" ], "unverified": [] }, { "id": "laion-clip-vit-bigg-14-laion2b-39b-b160k", "name": "OpenCLIP ViT-bigG-14 LAION-2B", "level": "variant", "family_id": "laion-openclip", "developer": [ "laion" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-01-23", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenCLIP ViT-bigG-14 LAION-2B, a checkpoint of the OpenCLIP (LAION) family. Open reproduction of CLIP trained on the public LAION-2B image-text dataset, used for zero-shot classification and retrieval and as the text encoder of image generators.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2212.07143", "https://huggingface.co/laion/CLIP-ViT-bigG-14-laion2B-39B-b160k", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://modelzoo.co/model/open-clip" ], "unverified": [] }, { "id": "laion-clap-htsat-fused", "name": "CLAP (HTSAT fused)", "level": "model", "family_id": null, "developer": [ "laion" ], "model_class": "embedding", "modalities_in": [ "text", "audio" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 153558316, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-02-16", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Contrastive language-audio model that embeds sounds and text in one space, for zero-shot audio classification and audio search.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/laion/clap-htsat-fused", "https://huggingface.co/api/models?pipeline_tag=audio-classification&sort=downloads&limit=50" ], "unverified": [] }, { "id": "google-siglip-2", "name": "SigLIP 2", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "ViT-B (86M) to g (1B); NaFlex versions keep native aspect ratio. The first SigLIP 400M encoder is a separate AURION record (siglip-400m).", "context_length": null, "release_date": "2025-02-17", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multilingual image-text encoders trained with a sigmoid loss plus captioning and self-supervised objectives, for zero-shot classification, retrieval and as VLM vision encoders.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2502.14786", "https://github.com/google-research/big_vision", "https://aihub.qualcomm.com/models/siglip2", "https://huggingface.co/api/models?pipeline_tag=zero-shot-image-classification&sort=downloads&limit=50", "https://huggingface.co/google/siglip2-base-patch16-224", "https://huggingface.co/google/siglip2-so400m-patch14-384" ], "unverified": [] }, { "id": "google-siglip2-base-patch16-224", "name": "SigLIP 2 base patch16 224", "level": "variant", "family_id": "google-siglip-2", "developer": [ "google" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 375187970, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-02-17", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "SigLIP 2 base patch16 224, a checkpoint of the SigLIP 2 family. Multilingual image-text encoders trained with a sigmoid loss plus captioning and self-supervised objectives, for zero-shot classification, retrieval and as VLM vision encoders.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2502.14786", "https://huggingface.co/google/siglip2-base-patch16-224", "https://aihub.qualcomm.com/models/siglip2", "https://huggingface.co/api/models?pipeline_tag=zero-shot-image-classification&sort=downloads&limit=50" ], "unverified": [] }, { "id": "google-siglip2-so400m-patch14-384", "name": "SigLIP 2 So400m patch14 384", "level": "variant", "family_id": "google-siglip-2", "developer": [ "google" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 1136008498, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-02-17", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "SigLIP 2 So400m patch14 384, a checkpoint of the SigLIP 2 family. Multilingual image-text encoders trained with a sigmoid loss plus captioning and self-supervised objectives, for zero-shot classification, retrieval and as VLM vision encoders.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2502.14786", "https://huggingface.co/google/siglip2-so400m-patch14-384", "https://aihub.qualcomm.com/models/siglip2", "https://huggingface.co/api/models?pipeline_tag=zero-shot-image-classification&sort=downloads&limit=50" ], "unverified": [] }, { "id": "google-siglip2-giant-opt-patch16-384", "name": "SigLIP 2 giant patch16 384", "level": "variant", "family_id": "google-siglip-2", "developer": [ "google" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 1871885426, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-02-17", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "SigLIP 2 giant patch16 384, a checkpoint of the SigLIP 2 family. Multilingual image-text encoders trained with a sigmoid loss plus captioning and self-supervised objectives, for zero-shot classification, retrieval and as VLM vision encoders.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2502.14786", "https://huggingface.co/google/siglip2-giant-opt-patch16-384", "https://aihub.qualcomm.com/models/siglip2", "https://huggingface.co/api/models?pipeline_tag=zero-shot-image-classification&sort=downloads&limit=50" ], "unverified": [] }, { "id": "baai-eva-clip", "name": "EVA-CLIP", "level": "model", "family_id": null, "developer": [ "baai" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-03-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "CLIP models trained with improved recipes from EVA-initialised vision encoders, reaching higher zero-shot accuracy at lower training cost.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2303.15389", "https://github.com/baaivision/EVA/tree/master/EVA-CLIP", "https://huggingface.co/QuanSun/EVA-CLIP", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md" ], "unverified": [] }, { "id": "meta-metaclip-2-worldwide-huge-quickgelu", "name": "MetaCLIP 2 (worldwide, ViT-H)", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 1858783745, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-07-31", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "CLIP trained from scratch on worldwide multilingual image-text pairs with Meta's curation recipe, for multilingual image-text retrieval.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/facebook/metaclip-2-worldwide-huge-quickgelu", "https://keras.io/keras_hub/presets/", "https://www.kaggle.com/models/keras/metaclip2" ], "unverified": [] }, { "id": "apple-mobileclip", "name": "MobileCLIP", "level": "family", "family_id": null, "developer": [ "apple" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Hybrid convolution and transformer image encoders", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-11-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "apple-amlr", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fast image-text models for phones, trained with multi-modal reinforced training, for on-device zero-shot classification and search.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2311.17049", "https://github.com/apple/ml-mobileclip", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://huggingface.co/apple/MobileCLIP-S0", "https://huggingface.co/apple/MobileCLIP2-S0" ], "unverified": [] }, { "id": "apple-mobileclip-s0", "name": "MobileCLIP-S0", "level": "variant", "family_id": "apple-mobileclip", "developer": [ "apple" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-03-06", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "apple-amlr", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MobileCLIP-S0, a checkpoint of the MobileCLIP family. Fast image-text models for phones, trained with multi-modal reinforced training, for on-device zero-shot classification and search.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2311.17049", "https://huggingface.co/apple/MobileCLIP-S0", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md" ], "unverified": [] }, { "id": "apple-mobileclip2-s0", "name": "MobileCLIP2-S0", "level": "variant", "family_id": "apple-mobileclip", "developer": [ "apple" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-08-25", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "apple-amlr", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MobileCLIP2-S0, a checkpoint of the MobileCLIP family. Fast image-text models for phones, trained with multi-modal reinforced training, for on-device zero-shot classification and search.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2311.17049", "https://huggingface.co/apple/MobileCLIP2-S0", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://huggingface.co/apple/MobileCLIP-S0" ], "unverified": [] }, { "id": "apple-dfn5b-clip-vit-h-14-378", "name": "DFN5B-CLIP ViT-H-14-378", "level": "model", "family_id": null, "developer": [ "apple" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-09-29", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "apple-amlr", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "CLIP model trained on 5 billion image-text pairs selected by a data filtering network, for zero-shot classification and retrieval.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2309.17425", "https://huggingface.co/apple/DFN5B-CLIP-ViT-H-14-378", "https://huggingface.co/apple/DFN5B-CLIP-ViT-H-14" ], "unverified": [] }, { "id": "alibaba-damo-chinese-clip-vit-base-patch16", "name": "Chinese-CLIP ViT-B/16", "level": "model", "family_id": null, "developer": [ "alibaba-damo" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-11-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Chinese", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "CLIP trained on about 200 million Chinese image-text pairs, for Chinese image-text retrieval and zero-shot classification.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2211.01335", "https://huggingface.co/OFA-Sys/chinese-clip-vit-base-patch16", "https://huggingface.co/api/models?pipeline_tag=zero-shot-image-classification&sort=downloads&limit=50", "https://www.modelscope.cn/models/iic/multi-modal_clip-vit-large-patch14_336_zh" ], "unverified": [] }, { "id": "microsoft-biomedclip", "name": "BiomedCLIP", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-03-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Biomedical image-text model trained on 15 million figure-caption pairs from PubMed Central, for medical image retrieval and zero-shot classification.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2303.00915", "https://huggingface.co/microsoft/BiomedCLIP-PubMedBERT_256-vit_base_patch16_224", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://ai.azure.com/catalog/models?industry=health-and-life-sciences" ], "unverified": [] }, { "id": "meta-imagebind", "name": "ImageBind", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "embedding", "modalities_in": [ "text", "image", "video", "audio", "sensor" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-05-09", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-sa-4-0", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "One embedding space for six modalities (images, text, audio, depth, thermal and motion sensor data), learned by binding each to images; non-commercial licence.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2305.05665", "https://github.com/facebookresearch/ImageBind", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://replicate.com/daanelson/imagebind" ], "unverified": [] }, { "id": "meta-perception-encoder", "name": "Perception Encoder", "level": "family", "family_id": null, "developer": [ "meta" ], "model_class": "embedding", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-04-11", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vision encoders trained with contrastive image and video-text learning, with spatial and language-aligned versions for retrieval, detection and VLMs.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2504.13181", "https://github.com/facebookresearch/perception_models", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://docs.roboflow.com/models/supported-models/perception-encoder.md", "https://huggingface.co/facebook/PE-Core-L14-336", "https://huggingface.co/facebook/PE-Core-G14-448" ], "unverified": [] }, { "id": "meta-pe-core-l14-336", "name": "PE-Core-L14-336", "level": "variant", "family_id": "meta-perception-encoder", "developer": [ "meta" ], "model_class": "embedding", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-04-11", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "PE-Core-L14-336, a checkpoint of the Perception Encoder family. Vision encoders trained with contrastive image and video-text learning, with spatial and language-aligned versions for retrieval, detection and VLMs.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2504.13181", "https://huggingface.co/facebook/PE-Core-L14-336", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://docs.roboflow.com/models/supported-models/perception-encoder.md" ], "unverified": [] }, { "id": "meta-pe-core-g14-448", "name": "PE-Core-G14-448", "level": "variant", "family_id": "meta-perception-encoder", "developer": [ "meta" ], "model_class": "embedding", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-04-11", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "PE-Core-G14-448, a checkpoint of the Perception Encoder family. Vision encoders trained with contrastive image and video-text learning, with spatial and language-aligned versions for retrieval, detection and VLMs.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2504.13181", "https://huggingface.co/facebook/PE-Core-G14-448", "https://github.com/huggingface/pytorch-image-models/blob/main/README.md", "https://docs.roboflow.com/models/supported-models/perception-encoder.md" ], "unverified": [] }, { "id": "microsoft-xclip-base-patch32", "name": "X-CLIP base patch32", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "embedding", "modalities_in": [ "text", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 196585855, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2022-08-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "CLIP extended to video with cross-frame attention, for zero-shot and few-shot video classification and video-text retrieval.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2208.02816", "https://huggingface.co/microsoft/xclip-base-patch32", "https://huggingface.co/api/models?pipeline_tag=video-classification&sort=downloads&limit=50" ], "unverified": [] }, { "id": "microsoft-florence-2", "name": "Florence-2", "level": "family", "family_id": null, "developer": [ "microsoft" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-11-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small sequence-to-sequence vision foundation model that handles captioning, object detection, grounding, segmentation and OCR from text prompts.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2311.06242", "https://huggingface.co/api/models?pipeline_tag=image-text-to-text&sort=downloads&limit=50", "https://replicate.com/lucataco/florence-2-large", "https://huggingface.co/microsoft/Florence-2-base", "https://huggingface.co/microsoft/Florence-2-large" ], "unverified": [] }, { "id": "microsoft-florence-2-base", "name": "Florence-2 base", "level": "variant", "family_id": "microsoft-florence-2", "developer": [ "microsoft" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 231567705, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-06-15", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Florence-2 base, a checkpoint of the Florence-2 family. Small sequence-to-sequence vision foundation model that handles captioning, object detection, grounding, segmentation and OCR from text prompts.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2311.06242", "https://huggingface.co/microsoft/Florence-2-base", "https://huggingface.co/api/models?pipeline_tag=image-text-to-text&sort=downloads&limit=50", "https://replicate.com/lucataco/florence-2-large" ], "unverified": [] }, { "id": "microsoft-florence-2-large", "name": "Florence-2 large", "level": "variant", "family_id": "microsoft-florence-2", "developer": [ "microsoft" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 776721497, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-06-15", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Florence-2 large, a checkpoint of the Florence-2 family. Small sequence-to-sequence vision foundation model that handles captioning, object detection, grounding, segmentation and OCR from text prompts.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2311.06242", "https://huggingface.co/microsoft/Florence-2-large", "https://huggingface.co/api/models?pipeline_tag=image-text-to-text&sort=downloads&limit=50", "https://replicate.com/lucataco/florence-2-large" ], "unverified": [] }, { "id": "microsoft-kosmos-2", "name": "Kosmos-2", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 1664485376, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-10-02", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multimodal language model that links phrases in its text to boxes in the image, for grounded captioning and referring expressions.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/microsoft/kosmos-2-patch14-224", "https://huggingface.co/api/models?pipeline_tag=image-to-text&sort=downloads&limit=50" ], "unverified": [] }, { "id": "llava-llava", "name": "LLaVA", "level": "family", "family_id": null, "developer": [ "university-of-wisconsin-madison", "microsoft" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "LLaVA, LLaVA-1.5, LLaVA-NeXT (1.6), NeXT-Video and OneVision; later versions involve ByteDance.", "context_length": null, "release_date": "2023-04-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, zh", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Visual instruction tuning: a vision encoder connected to an open LLM and trained on GPT-generated image instruction data, the template for many open VLMs.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2304.08485", "https://llava-vl.github.io/", "https://github.com/haotian-liu/LLaVA", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/llava", "https://huggingface.co/llava-hf/llava-1.5-7b-hf", "https://huggingface.co/llava-hf/llava-v1.6-mistral-7b-hf" ], "unverified": [] }, { "id": "llava-llava-1-5-7b", "name": "LLaVA-1.5 7B", "level": "variant", "family_id": "llava-llava", "developer": [ "university-of-wisconsin-madison", "microsoft" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 7063427072, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-12-05", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "llama-2-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "LLaVA-1.5 7B, a checkpoint of the LLaVA family. Visual instruction tuning: a vision encoder connected to an open LLM and trained on GPT-generated image instruction data, the template for many open VLMs.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2304.08485", "https://huggingface.co/llava-hf/llava-1.5-7b-hf", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/llava" ], "unverified": [] }, { "id": "llava-llava-v1-6-mistral-7b", "name": "LLaVA-NeXT (1.6) Mistral 7B", "level": "variant", "family_id": "llava-llava", "developer": [ "university-of-wisconsin-madison", "microsoft" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 7566747648, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-02-20", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "LLaVA-NeXT (1.6) Mistral 7B, a checkpoint of the LLaVA family. Visual instruction tuning: a vision encoder connected to an open LLM and trained on GPT-generated image instruction data, the template for many open VLMs.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2304.08485", "https://huggingface.co/llava-hf/llava-v1.6-mistral-7b-hf", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/llava" ], "unverified": [] }, { "id": "llava-llava-next-video-7b", "name": "LLaVA-NeXT-Video 7B", "level": "variant", "family_id": "llava-llava", "developer": [ "university-of-wisconsin-madison", "microsoft" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 7063431168, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-06-05", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "llama-2-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "LLaVA-NeXT-Video 7B, a checkpoint of the LLaVA family. Visual instruction tuning: a vision encoder connected to an open LLM and trained on GPT-generated image instruction data, the template for many open VLMs.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2304.08485", "https://huggingface.co/llava-hf/LLaVA-NeXT-Video-7B-hf", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/llava" ], "unverified": [] }, { "id": "llava-llava-onevision-qwen2-7b-ov", "name": "LLaVA-OneVision Qwen2 7B", "level": "variant", "family_id": "llava-llava", "developer": [ "bytedance" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8030807584, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-08-13", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, zh", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "LLaVA-OneVision Qwen2 7B, a checkpoint of the LLaVA family. Visual instruction tuning: a vision encoder connected to an open LLM and trained on GPT-generated image instruction data, the template for many open VLMs.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2304.08485", "https://huggingface.co/llava-hf/llava-onevision-qwen2-7b-ov-hf", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/llava" ], "unverified": [ "developer" ] }, { "id": "microsoft-llava-med-v1-5-mistral-7b", "name": "LLaVA-Med v1.5 (Mistral 7B)", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 7566219264, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-05-14", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "LLaVA adapted to biomedicine with PubMed Central figure-caption data, for research on biomedical visual question answering.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/microsoft/llava-med-v1.5-mistral-7b", "https://ai.azure.com/catalog/models?industry=health-and-life-sciences" ], "unverified": [] }, { "id": "salesforce-blip", "name": "BLIP", "level": "family", "family_id": null, "developer": [ "salesforce" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-01-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "bsd-3-clause", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vision-language pre-training that bootstraps captions from noisy web data, with models for image captioning, visual question answering and image-text retrieval.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2201.12086", "https://github.com/salesforce/BLIP", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://paperswithcode.com/method/blip", "https://huggingface.co/Salesforce/blip-image-captioning-large", "https://huggingface.co/Salesforce/blip-vqa-base" ], "unverified": [] }, { "id": "salesforce-blip-image-captioning-large", "name": "BLIP image captioning large", "level": "variant", "family_id": "salesforce-blip", "developer": [ "salesforce" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 469733436, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2022-12-13", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "bsd-3-clause", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BLIP image captioning large, a checkpoint of the BLIP family. Vision-language pre-training that bootstraps captions from noisy web data, with models for image captioning, visual question answering and image-text retrieval.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2201.12086", "https://huggingface.co/Salesforce/blip-image-captioning-large", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://paperswithcode.com/method/blip" ], "unverified": [] }, { "id": "salesforce-blip-vqa-base", "name": "BLIP VQA base", "level": "variant", "family_id": "salesforce-blip", "developer": [ "salesforce" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 384673596, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2022-12-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "bsd-3-clause", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BLIP VQA base, a checkpoint of the BLIP family. Vision-language pre-training that bootstraps captions from noisy web data, with models for image captioning, visual question answering and image-text retrieval.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2201.12086", "https://huggingface.co/Salesforce/blip-vqa-base", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://paperswithcode.com/method/blip" ], "unverified": [] }, { "id": "salesforce-blip-2", "name": "BLIP-2", "level": "family", "family_id": null, "developer": [ "salesforce" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-01-30", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Connects a frozen image encoder to a frozen LLM through a small Querying Transformer, giving captioning and visual question answering at low training cost.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2301.12597", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://ai.azure.com/catalog/models?task=image-to-text", "https://huggingface.co/Salesforce/blip2-opt-2.7b", "https://huggingface.co/Salesforce/blip2-flan-t5-xl" ], "unverified": [] }, { "id": "salesforce-blip2-opt-2-7b", "name": "BLIP-2 OPT 2.7B", "level": "variant", "family_id": "salesforce-blip-2", "developer": [ "salesforce" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 3744761856, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-02-06", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BLIP-2 OPT 2.7B, a checkpoint of the BLIP-2 family. Connects a frozen image encoder to a frozen LLM through a small Querying Transformer, giving captioning and visual question answering at low training cost.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2301.12597", "https://huggingface.co/Salesforce/blip2-opt-2.7b", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://ai.azure.com/catalog/models?task=image-to-text" ], "unverified": [] }, { "id": "salesforce-blip2-flan-t5-xl", "name": "BLIP-2 Flan-T5 XL", "level": "variant", "family_id": "salesforce-blip-2", "developer": [ "salesforce" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 3942446592, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-02-06", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BLIP-2 Flan-T5 XL, a checkpoint of the BLIP-2 family. Connects a frozen image encoder to a frozen LLM through a small Querying Transformer, giving captioning and visual question answering at low training cost.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2301.12597", "https://huggingface.co/Salesforce/blip2-flan-t5-xl", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://ai.azure.com/catalog/models?task=image-to-text" ], "unverified": [] }, { "id": "salesforce-xgen-mm-phi3-mini-instruct", "name": "xGen-MM (BLIP-3) phi3 mini instruct", "level": "model", "family_id": null, "developer": [ "salesforce" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 4359257219, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-08-09", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small open multimodal model of the xGen-MM (BLIP-3) line that takes interleaved images and text.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Salesforce/xgen-mm-phi3-mini-instruct-interleave-r-v1.5", "https://replicate.com/salesforce/blip" ], "unverified": [] }, { "id": "hugging-face-idefics", "name": "Idefics", "level": "family", "family_id": null, "developer": [ "hugging-face" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-04-09", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Hugging Face open vision language models that take any sequence of images and text and answer, caption or read documents.", "verified_on": "2026-10-12", "sources": [ "https://ai.azure.com/catalog/models?task=image-text-to-text", "https://replicate.com/zsxkib/idefics3", "https://huggingface.co/HuggingFaceM4/idefics2-8b", "https://huggingface.co/HuggingFaceM4/Idefics3-8B-Llama3" ], "unverified": [] }, { "id": "hugging-face-idefics2-8b", "name": "Idefics2 8B", "level": "variant", "family_id": "hugging-face-idefics", "developer": [ "hugging-face" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8402768112, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-04-09", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Idefics2 8B, a checkpoint of the Idefics family. Hugging Face open vision language models that take any sequence of images and text and answer, caption or read documents.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/HuggingFaceM4/idefics2-8b", "https://ai.azure.com/catalog/models?task=image-text-to-text", "https://replicate.com/zsxkib/idefics3" ], "unverified": [] }, { "id": "hugging-face-idefics3-8b-llama3", "name": "Idefics3 8B Llama3", "level": "variant", "family_id": "hugging-face-idefics", "developer": [ "hugging-face" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8462086384, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-08-05", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Idefics3 8B Llama3, a checkpoint of the Idefics family. Hugging Face open vision language models that take any sequence of images and text and answer, caption or read documents.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/HuggingFaceM4/Idefics3-8B-Llama3", "https://ai.azure.com/catalog/models?task=image-text-to-text", "https://replicate.com/zsxkib/idefics3" ], "unverified": [] }, { "id": "hugging-face-smolvlm", "name": "SmolVLM", "level": "family", "family_id": null, "developer": [ "hugging-face" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-11-18", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "laptop", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small open vision language models (256M to 2.2B) for images and video that run on laptops and phones.", "verified_on": "2026-10-12", "sources": [ "https://replicate.com/lucataco/smolvlm-instruct", "https://docs.roboflow.com/models/supported-models/smolvlm2.md", "https://huggingface.co/HuggingFaceTB/SmolVLM-Instruct", "https://huggingface.co/HuggingFaceTB/SmolVLM2-2.2B-Instruct" ], "unverified": [] }, { "id": "hugging-face-smolvlm-instruct", "name": "SmolVLM Instruct", "level": "variant", "family_id": "hugging-face-smolvlm", "developer": [ "hugging-face" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 2246272880, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-11-18", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "SmolVLM Instruct, a checkpoint of the SmolVLM family. Small open vision language models (256M to 2.2B) for images and video that run on laptops and phones.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/HuggingFaceTB/SmolVLM-Instruct", "https://replicate.com/lucataco/smolvlm-instruct", "https://docs.roboflow.com/models/supported-models/smolvlm2.md" ], "unverified": [] }, { "id": "hugging-face-smolvlm2-2-2b-instruct", "name": "SmolVLM2 2.2B Instruct", "level": "variant", "family_id": "hugging-face-smolvlm", "developer": [ "hugging-face" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 2246784880, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-02-08", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "SmolVLM2 2.2B Instruct, a checkpoint of the SmolVLM family. Small open vision language models (256M to 2.2B) for images and video that run on laptops and phones.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/HuggingFaceTB/SmolVLM2-2.2B-Instruct", "https://replicate.com/lucataco/smolvlm-instruct", "https://docs.roboflow.com/models/supported-models/smolvlm2.md" ], "unverified": [] }, { "id": "hugging-face-smolvlm2-500m-video-instruct", "name": "SmolVLM2 500M Video Instruct", "level": "variant", "family_id": "hugging-face-smolvlm", "developer": [ "hugging-face" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 507482304, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-02-11", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "SmolVLM2 500M Video Instruct, a checkpoint of the SmolVLM family. Small open vision language models (256M to 2.2B) for images and video that run on laptops and phones.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/HuggingFaceTB/SmolVLM2-500M-Video-Instruct", "https://replicate.com/lucataco/smolvlm-instruct", "https://docs.roboflow.com/models/supported-models/smolvlm2.md" ], "unverified": [] }, { "id": "hugging-face-smolvlm2-256m-video-instruct", "name": "SmolVLM2 256M Video Instruct", "level": "variant", "family_id": "hugging-face-smolvlm", "developer": [ "hugging-face" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 256484928, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-02-11", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "SmolVLM2 256M Video Instruct, a checkpoint of the SmolVLM family. Small open vision language models (256M to 2.2B) for images and video that run on laptops and phones.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/HuggingFaceTB/SmolVLM2-256M-Video-Instruct", "https://replicate.com/lucataco/smolvlm-instruct", "https://docs.roboflow.com/models/supported-models/smolvlm2.md" ], "unverified": [] }, { "id": "naver-vilt-b32-finetuned-vqa", "name": "ViLT (VQA fine-tune)", "level": "model", "family_id": null, "developer": [ "naver" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-02-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vision-and-language transformer that takes image patches directly without a convolutional or region detector, here fine-tuned for visual question answering.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2102.03334", "https://huggingface.co/dandelin/vilt-b32-finetuned-vqa", "https://huggingface.co/api/models?pipeline_tag=visual-question-answering&sort=downloads&limit=50", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden" ], "unverified": [ "developer" ] }, { "id": "alibaba-damo-ofa", "name": "OFA", "level": "model", "family_id": null, "developer": [ "alibaba-damo" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text", "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-02-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Unified sequence-to-sequence model for image captioning, visual question answering, grounding and text-to-image tasks in one architecture.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2202.03052", "https://github.com/OFA-Sys/OFA", "https://huggingface.co/OFA-Sys/ofa-large", "https://www.modelscope.cn/models/iic/ofa_image-caption_coco_large_en", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/ofa" ], "unverified": [] }, { "id": "alibaba-damo-mplug-owl", "name": "mPLUG-Owl", "level": "family", "family_id": null, "developer": [ "alibaba-damo" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-04-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba multimodal LLMs that add a modular visual encoder to a language model; Owl3 handles long image sequences and video.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2304.14178", "https://github.com/X-PLUG/mPLUG-Owl", "https://www.modelscope.cn/models/iic/mPLUG-Owl3-7B-241101", "https://www.modelscope.cn/models/iic/mplug_image-captioning_coco_base_zh", "https://huggingface.co/mPLUG/mPLUG-Owl3-7B-240728" ], "unverified": [] }, { "id": "alibaba-damo-mplug-owl3-7b", "name": "mPLUG-Owl3 7B", "level": "variant", "family_id": "alibaba-damo-mplug-owl", "developer": [ "alibaba-damo" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8061197888, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-08-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "mPLUG-Owl3 7B, a checkpoint of the mPLUG-Owl family. Alibaba multimodal LLMs that add a modular visual encoder to a language model; Owl3 handles long image sequences and video.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2304.14178", "https://huggingface.co/mPLUG/mPLUG-Owl3-7B-240728", "https://www.modelscope.cn/models/iic/mPLUG-Owl3-7B-241101", "https://www.modelscope.cn/models/iic/mplug_image-captioning_coco_base_zh" ], "unverified": [] }, { "id": "alibaba-damo-videollama", "name": "VideoLLaMA", "level": "family", "family_id": null, "developer": [ "alibaba-damo" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-06-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Video understanding language models from DAMO Academy Singapore; VideoLLaMA 3 is image-centric and handles both images and long video.", "verified_on": "2026-10-12", "sources": [ "https://github.com/DAMO-NLP-SG/VideoLLaMA3", "https://huggingface.co/api/models?pipeline_tag=video-text-to-text&sort=downloads&limit=50", "https://replicate.com/lucataco/videollama3-7b", "https://huggingface.co/DAMO-NLP-SG/VideoLLaMA3-7B", "https://huggingface.co/DAMO-NLP-SG/VideoLLaMA2.1-7B-16F" ], "unverified": [] }, { "id": "alibaba-damo-videollama3-7b", "name": "VideoLLaMA3 7B", "level": "variant", "family_id": "alibaba-damo-videollama", "developer": [ "alibaba-damo" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8044744944, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-01-21", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "VideoLLaMA3 7B, a checkpoint of the VideoLLaMA family. Video understanding language models from DAMO Academy Singapore; VideoLLaMA 3 is image-centric and handles both images and long video.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/DAMO-NLP-SG/VideoLLaMA3-7B", "https://huggingface.co/api/models?pipeline_tag=video-text-to-text&sort=downloads&limit=50", "https://replicate.com/lucataco/videollama3-7b" ], "unverified": [] }, { "id": "alibaba-damo-videollama2-1-7b-16f", "name": "VideoLLaMA2.1 7B 16F", "level": "variant", "family_id": "alibaba-damo-videollama", "developer": [ "alibaba-damo" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8420731360, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-10-14", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "VideoLLaMA2.1 7B 16F, a checkpoint of the VideoLLaMA family. Video understanding language models from DAMO Academy Singapore; VideoLLaMA 3 is image-centric and handles both images and long video.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/DAMO-NLP-SG/VideoLLaMA2.1-7B-16F", "https://huggingface.co/api/models?pipeline_tag=video-text-to-text&sort=downloads&limit=50", "https://replicate.com/lucataco/videollama3-7b" ], "unverified": [] }, { "id": "alibaba-damo-sealmmm-7b-v0-1", "name": "SeaLMMM 7B v0.1", "level": "model", "family_id": null, "developer": [ "alibaba-damo" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 7687842816, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-04-22", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, zh, vi, id, th, ms, km, lo, my, tl", "indonesian_support": "listed by developer", "regional_languages": [ "Vietnamese", "Thai", "Malay", "Khmer", "Lao", "Burmese", "Tagalog" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Multimodal model from the SeaLLMs project that reads images and text in Southeast Asian languages as well as English and Chinese.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/SeaLLMs/SeaLMMM-7B-v0.1", "https://ai.azure.com/catalog/models?task=image-text-to-text" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "kaust-minigpt-4", "name": "MiniGPT-4", "level": "model", "family_id": null, "developer": [ "kaust" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-04-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Aligns a frozen visual encoder with the Vicuna LLM through one projection layer, showing detailed image description and image-grounded chat.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2304.10592", "https://github.com/Vision-CAIR/MiniGPT-4", "https://replicate.com/daanelson/minigpt-4", "https://github.com/open-mmlab/mmpretrain/tree/main/configs/minigpt4" ], "unverified": [] }, { "id": "meta-longvu", "name": "LongVU", "level": "model", "family_id": null, "developer": [ "meta", "kaust" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 7671665664, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-10-18", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Video language model that adaptively compresses frames in space and time so it can understand long videos within a limited context.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Vision-CAIR/LongVU_Qwen2_7B", "https://huggingface.co/api/models?pipeline_tag=video-text-to-text&sort=likes&limit=50" ], "unverified": [] }, { "id": "adept-fuyu-8b", "name": "Fuyu-8B", "level": "model", "family_id": null, "developer": [ "adept" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 9408238592, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-10-17", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Decoder-only multimodal model that feeds image patches straight into the transformer without a separate image encoder, for charts, diagrams and screen reading.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/adept/fuyu-8b", "https://replicate.com/lucataco/fuyu-8b" ], "unverified": [] }, { "id": "01-ai-yi-1-5", "name": "Yi-1.5", "level": "family", "family_id": null, "developer": [ "01-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-05-10", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Upgraded Yi models continuously pre-trained on 500 billion more tokens, stronger in coding, maths and reasoning, in English and Chinese.", "verified_on": "2026-10-12", "sources": [ "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html", "https://huggingface.co/01-ai/Yi-1.5-9B-Chat", "https://huggingface.co/01-ai/Yi-1.5-34B-Chat" ], "unverified": [] }, { "id": "01-ai-yi-1-5-9b-chat", "name": "Yi-1.5 9B Chat", "level": "variant", "family_id": "01-ai-yi-1-5", "developer": [ "01-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8829407232, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-05-10", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Yi-1.5 9B Chat, a checkpoint of the Yi-1.5 family. Upgraded Yi models continuously pre-trained on 500 billion more tokens, stronger in coding, maths and reasoning, in English and Chinese.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/01-ai/Yi-1.5-9B-Chat", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html" ], "unverified": [] }, { "id": "01-ai-yi-1-5-34b-chat", "name": "Yi-1.5 34B Chat", "level": "variant", "family_id": "01-ai-yi-1-5", "developer": [ "01-ai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 34388917248, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-05-10", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Yi-1.5 34B Chat, a checkpoint of the Yi-1.5 family. Upgraded Yi models continuously pre-trained on 500 billion more tokens, stronger in coding, maths and reasoning, in English and Chinese.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/01-ai/Yi-1.5-34B-Chat", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html" ], "unverified": [] }, { "id": "01-ai-yi-vl", "name": "Yi-VL", "level": "family", "family_id": null, "developer": [ "01-ai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-12-25", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open vision language models built on Yi language models, for image understanding and multi-round image chat in English and Chinese.", "verified_on": "2026-10-12", "sources": [ "https://www.modelscope.cn/models/01ai/Yi-VL-34B", "https://huggingface.co/01-ai/Yi-VL-6B", "https://huggingface.co/01-ai/Yi-VL-34B" ], "unverified": [] }, { "id": "01-ai-yi-vl-6b", "name": "Yi-VL 6B", "level": "variant", "family_id": "01-ai-yi-vl", "developer": [ "01-ai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-12-25", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Yi-VL 6B, a checkpoint of the Yi-VL family. Open vision language models built on Yi language models, for image understanding and multi-round image chat in English and Chinese.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/01-ai/Yi-VL-6B", "https://www.modelscope.cn/models/01ai/Yi-VL-34B" ], "unverified": [] }, { "id": "01-ai-yi-vl-34b", "name": "Yi-VL 34B", "level": "variant", "family_id": "01-ai-yi-vl", "developer": [ "01-ai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-12-25", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Yi-VL 34B, a checkpoint of the Yi-VL family. Open vision language models built on Yi language models, for image understanding and multi-round image chat in English and Chinese.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/01-ai/Yi-VL-34B", "https://www.modelscope.cn/models/01ai/Yi-VL-34B" ], "unverified": [] }, { "id": "zhipu-zai-chatglm", "name": "ChatGLM", "level": "family", "family_id": null, "developer": [ "zhipu-zai", "tsinghua-university" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-06-24", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: zh, en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open bilingual Chinese-English chat models on the GLM architecture, sized to run on a single consumer GPU.", "verified_on": "2026-10-12", "sources": [ "https://www.kaggle.com/models/thudm-keg/chatglm2", "https://catalog.ngc.nvidia.com/orgs/nvidia/models/chatglm3-6b-chat-int4", "https://huggingface.co/zai-org/chatglm2-6b", "https://huggingface.co/zai-org/chatglm3-6b" ], "unverified": [ "release_date" ] }, { "id": "zhipu-zai-chatglm2-6b", "name": "ChatGLM2-6B", "level": "variant", "family_id": "zhipu-zai-chatglm", "developer": [ "zhipu-zai", "tsinghua-university" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-06-24", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: zh, en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ChatGLM2-6B, a checkpoint of the ChatGLM family. Open bilingual Chinese-English chat models on the GLM architecture, sized to run on a single consumer GPU.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/zai-org/chatglm2-6b", "https://www.kaggle.com/models/thudm-keg/chatglm2", "https://catalog.ngc.nvidia.com/orgs/nvidia/models/chatglm3-6b-chat-int4" ], "unverified": [] }, { "id": "zhipu-zai-chatglm3-6b", "name": "ChatGLM3-6B", "level": "variant", "family_id": "zhipu-zai-chatglm", "developer": [ "zhipu-zai", "tsinghua-university" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 6243584032, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-10-25", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: zh, en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ChatGLM3-6B, a checkpoint of the ChatGLM family. Open bilingual Chinese-English chat models on the GLM architecture, sized to run on a single consumer GPU.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/zai-org/chatglm3-6b", "https://www.kaggle.com/models/thudm-keg/chatglm2", "https://catalog.ngc.nvidia.com/orgs/nvidia/models/chatglm3-6b-chat-int4" ], "unverified": [] }, { "id": "zhipu-zai-glm-4-9b-chat", "name": "GLM-4-9B-Chat", "level": "model", "family_id": null, "developer": [ "zhipu-zai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 9399951392, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-06-04", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: zh, en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open 9B chat model of the GLM-4 generation (June 2024) with tool use and 128k context; a 1M-context version exists. The later GLM-4-9B-0414 is a separate AURION record.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/zai-org/glm-4-9b-chat", "https://www.modelscope.cn/models/ZhipuAI/glm-4-9b-chat-1m" ], "unverified": [] }, { "id": "zhipu-zai-glm-4v-9b", "name": "GLM-4V-9B", "level": "model", "family_id": null, "developer": [ "zhipu-zai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 13906320928, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-06-04", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: zh, en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open 9B multimodal model of the GLM-4 generation for image understanding and chat in Chinese and English at 1120x1120 resolution.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/zai-org/glm-4v-9b", "https://www.modelscope.cn/models/ZhipuAI/glm-4v-9b" ], "unverified": [] }, { "id": "zhipu-zai-cogvlm", "name": "CogVLM", "level": "family", "family_id": null, "developer": [ "zhipu-zai", "tsinghua-university" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-10-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open visual language models that add a trainable visual expert module inside the language model; CogAgent specialises in GUI understanding for agents.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2311.03079", "https://github.com/THUDM/CogVLM", "https://www.modelscope.cn/models/ZhipuAI/cogvlm-chat", "https://huggingface.co/zai-org/cogvlm-chat-hf", "https://huggingface.co/zai-org/cogagent-chat-hf" ], "unverified": [] }, { "id": "zhipu-zai-cogvlm-chat-hf", "name": "CogVLM chat", "level": "variant", "family_id": "zhipu-zai-cogvlm", "developer": [ "zhipu-zai", "tsinghua-university" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 17639687424, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-11-16", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "CogVLM chat, a checkpoint of the CogVLM family. Open visual language models that add a trainable visual expert module inside the language model; CogAgent specialises in GUI understanding for agents.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2311.03079", "https://huggingface.co/zai-org/cogvlm-chat-hf", "https://www.modelscope.cn/models/ZhipuAI/cogvlm-chat" ], "unverified": [] }, { "id": "zhipu-zai-cogagent-chat-hf", "name": "CogAgent chat", "level": "variant", "family_id": "zhipu-zai-cogvlm", "developer": [ "zhipu-zai", "tsinghua-university" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 18291854528, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-12-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "CogAgent chat, a checkpoint of the CogVLM family. Open visual language models that add a trainable visual expert module inside the language model; CogAgent specialises in GUI understanding for agents.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2312.08914", "https://huggingface.co/zai-org/cogagent-chat-hf", "https://www.modelscope.cn/models/ZhipuAI/cogvlm-chat" ], "unverified": [] }, { "id": "shanghai-ai-lab-internlm", "name": "InternLM", "level": "family", "family_id": null, "developer": [ "shanghai-ai-lab" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-01-10", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open language models from Shanghai AI Laboratory, generations 2 to 3, with long context and strong reasoning in Chinese and English.", "verified_on": "2026-10-12", "sources": [ "https://github.com/InternLM/InternLM", "https://www.kaggle.com/models/intern-ai/internlm2", "https://www.kaggle.com/models/intern-ai/internlm2.5", "https://huggingface.co/internlm/internlm2-chat-7b", "https://huggingface.co/internlm/internlm2_5-7b-chat" ], "unverified": [] }, { "id": "shanghai-ai-lab-internlm2-chat-7b", "name": "InternLM2 chat 7B", "level": "variant", "family_id": "shanghai-ai-lab-internlm", "developer": [ "shanghai-ai-lab" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 7737708544, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-01-10", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "InternLM2 chat 7B, a checkpoint of the InternLM family. Open language models from Shanghai AI Laboratory, generations 2 to 3, with long context and strong reasoning in Chinese and English.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/internlm/internlm2-chat-7b", "https://www.kaggle.com/models/intern-ai/internlm2", "https://www.kaggle.com/models/intern-ai/internlm2.5" ], "unverified": [] }, { "id": "shanghai-ai-lab-internlm2-5-7b-chat", "name": "InternLM2.5 7B chat", "level": "variant", "family_id": "shanghai-ai-lab-internlm", "developer": [ "shanghai-ai-lab" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 7737708544, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-06-27", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "InternLM2.5 7B chat, a checkpoint of the InternLM family. Open language models from Shanghai AI Laboratory, generations 2 to 3, with long context and strong reasoning in Chinese and English.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/internlm/internlm2_5-7b-chat", "https://www.kaggle.com/models/intern-ai/internlm2", "https://www.kaggle.com/models/intern-ai/internlm2.5" ], "unverified": [] }, { "id": "shanghai-ai-lab-internlm3-8b-instruct", "name": "InternLM3 8B instruct", "level": "variant", "family_id": "shanghai-ai-lab-internlm", "developer": [ "shanghai-ai-lab" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8804241408, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-01-13", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "InternLM3 8B instruct, a checkpoint of the InternLM family. Open language models from Shanghai AI Laboratory, generations 2 to 3, with long context and strong reasoning in Chinese and English.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/internlm/internlm3-8b-instruct", "https://www.kaggle.com/models/intern-ai/internlm2", "https://www.kaggle.com/models/intern-ai/internlm2.5" ], "unverified": [] }, { "id": "shanghai-ai-lab-internlm-xcomposer2d5-7b", "name": "InternLM-XComposer 2.5 7B", "level": "model", "family_id": null, "developer": [ "shanghai-ai-lab" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-07-02", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vision language model for long-context image and video understanding and for composing illustrated articles and web pages.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/internlm/internlm-xcomposer2d5-7b", "https://huggingface.co/api/models?pipeline_tag=visual-question-answering&sort=likes&limit=50", "https://www.modelscope.cn/models/Shanghai_AI_Laboratory/internlm-xcomposer2d5-7b" ], "unverified": [] }, { "id": "shanghai-ai-lab-internvl", "name": "InternVL", "level": "family", "family_id": null, "developer": [ "shanghai-ai-lab" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-12-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open multimodal models from OpenGVLab that pair a large InternViT vision encoder with open LLMs, from 1B to over 200B parameters, across versions 1 to 3.5.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2312.14238", "https://github.com/OpenGVLab/InternVL", "https://www.modelscope.cn/models/OpenGVLab/InternVL-Chat-ViT-6B-Vicuna-13B-448px", "https://huggingface.co/api/models?pipeline_tag=reinforcement-learning&sort=downloads&limit=50", "https://huggingface.co/OpenGVLab/InternVL2-8B", "https://huggingface.co/OpenGVLab/InternVL3_5-2B" ], "unverified": [] }, { "id": "shanghai-ai-lab-internvl2-8b", "name": "InternVL2 8B", "level": "variant", "family_id": "shanghai-ai-lab-internvl", "developer": [ "shanghai-ai-lab" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8075365376, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-06-27", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "InternVL2 8B, a checkpoint of the InternVL family. Open multimodal models from OpenGVLab that pair a large InternViT vision encoder with open LLMs, from 1B to over 200B parameters, across versions 1 to 3.5.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2312.14238", "https://huggingface.co/OpenGVLab/InternVL2-8B", "https://www.modelscope.cn/models/OpenGVLab/InternVL-Chat-ViT-6B-Vicuna-13B-448px", "https://huggingface.co/api/models?pipeline_tag=reinforcement-learning&sort=downloads&limit=50" ], "unverified": [] }, { "id": "shanghai-ai-lab-internvl3-5-2b", "name": "InternVL3.5 2B", "level": "variant", "family_id": "shanghai-ai-lab-internvl", "developer": [ "shanghai-ai-lab" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 2348347392, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-08-25", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "InternVL3.5 2B, a checkpoint of the InternVL family. Open multimodal models from OpenGVLab that pair a large InternViT vision encoder with open LLMs, from 1B to over 200B parameters, across versions 1 to 3.5.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2312.14238", "https://huggingface.co/OpenGVLab/InternVL3_5-2B", "https://www.modelscope.cn/models/OpenGVLab/InternVL-Chat-ViT-6B-Vicuna-13B-448px", "https://huggingface.co/api/models?pipeline_tag=reinforcement-learning&sort=downloads&limit=50" ], "unverified": [] }, { "id": "shanghai-ai-lab-internvl3-5-8b", "name": "InternVL3.5 8B", "level": "variant", "family_id": "shanghai-ai-lab-internvl", "developer": [ "shanghai-ai-lab" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8528318464, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-08-25", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "InternVL3.5 8B, a checkpoint of the InternVL family. Open multimodal models from OpenGVLab that pair a large InternViT vision encoder with open LLMs, from 1B to over 200B parameters, across versions 1 to 3.5.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2312.14238", "https://huggingface.co/OpenGVLab/InternVL3_5-8B", "https://www.modelscope.cn/models/OpenGVLab/InternVL-Chat-ViT-6B-Vicuna-13B-448px", "https://huggingface.co/api/models?pipeline_tag=reinforcement-learning&sort=downloads&limit=50" ], "unverified": [] }, { "id": "shanghai-ai-lab-internvl3-5-241b-a28b", "name": "InternVL3.5 241B A28B", "level": "variant", "family_id": "shanghai-ai-lab-internvl", "developer": [ "shanghai-ai-lab" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 240699370368, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-08-25", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "InternVL3.5 241B A28B, a checkpoint of the InternVL family. Open multimodal models from OpenGVLab that pair a large InternViT vision encoder with open LLMs, from 1B to over 200B parameters, across versions 1 to 3.5.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2312.14238", "https://huggingface.co/OpenGVLab/InternVL3_5-241B-A28B", "https://www.modelscope.cn/models/OpenGVLab/InternVL-Chat-ViT-6B-Vicuna-13B-448px", "https://huggingface.co/api/models?pipeline_tag=reinforcement-learning&sort=downloads&limit=50" ], "unverified": [] }, { "id": "shanghai-ai-lab-internvideo2-5-chat-8b", "name": "InternVideo2.5 Chat 8B", "level": "model", "family_id": null, "developer": [ "shanghai-ai-lab" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8400921450, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-01-22", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Video multimodal model built on InternVL2.5 with long and rich context modelling, for fine-grained understanding of long videos.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/OpenGVLab/InternVideo2_5_Chat_8B", "https://huggingface.co/api/models?pipeline_tag=video-text-to-text&sort=downloads&limit=50" ], "unverified": [] }, { "id": "openbmb-minicpm-v", "name": "MiniCPM-V", "level": "family", "family_id": null, "developer": [ "openbmb" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-02-01", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Efficient open vision language models sized for phones and PCs, for OCR, document and video understanding; MiniCPM-V 4.6 is already an AURION record.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2408.01800", "https://github.com/OpenBMB/MiniCPM-o", "https://huggingface.co/api/models?pipeline_tag=visual-question-answering&sort=downloads&limit=50", "https://www.modelscope.cn/models/OpenBMB/MiniCPM-Llama3-V-2_5-int4", "https://huggingface.co/openbmb/MiniCPM-Llama3-V-2_5", "https://huggingface.co/openbmb/MiniCPM-V-2_6", "https://github.com/OpenBMB/MiniCPM-V" ], "unverified": [] }, { "id": "openbmb-minicpm-llama3-v-2-5", "name": "MiniCPM-Llama3-V 2.5", "level": "variant", "family_id": "openbmb-minicpm-v", "developer": [ "openbmb" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8537092336, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-05-19", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MiniCPM-Llama3-V 2.5, a checkpoint of the MiniCPM-V family. Efficient open vision language models sized for phones and PCs, for OCR, document and video understanding; MiniCPM-V 4.6 is already an AURION record.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2408.01800", "https://huggingface.co/openbmb/MiniCPM-Llama3-V-2_5", "https://huggingface.co/api/models?pipeline_tag=visual-question-answering&sort=downloads&limit=50", "https://www.modelscope.cn/models/OpenBMB/MiniCPM-Llama3-V-2_5-int4" ], "unverified": [] }, { "id": "openbmb-minicpm-v-2-6", "name": "MiniCPM-V 2.6", "level": "variant", "family_id": "openbmb-minicpm-v", "developer": [ "openbmb" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8099175152, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-08-04", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MiniCPM-V 2.6, a checkpoint of the MiniCPM-V family. Efficient open vision language models sized for phones and PCs, for OCR, document and video understanding; MiniCPM-V 4.6 is already an AURION record.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2408.01800", "https://huggingface.co/openbmb/MiniCPM-V-2_6", "https://huggingface.co/api/models?pipeline_tag=visual-question-answering&sort=downloads&limit=50", "https://www.modelscope.cn/models/OpenBMB/MiniCPM-Llama3-V-2_5-int4" ], "unverified": [] }, { "id": "openbmb-minicpm-v-4-5", "name": "MiniCPM-V 4.5", "level": "variant", "family_id": "openbmb-minicpm-v", "developer": [ "openbmb" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8695895280, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-08-24", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MiniCPM-V 4.5, a checkpoint of the MiniCPM-V family. Efficient open vision language models sized for phones and PCs, for OCR, document and video understanding; MiniCPM-V 4.6 is already an AURION record.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2408.01800", "https://huggingface.co/openbmb/MiniCPM-V-4_5", "https://huggingface.co/api/models?pipeline_tag=visual-question-answering&sort=downloads&limit=50", "https://www.modelscope.cn/models/OpenBMB/MiniCPM-Llama3-V-2_5-int4" ], "unverified": [] }, { "id": "openbmb-minicpm-o", "name": "MiniCPM-o", "level": "family", "family_id": null, "developer": [ "openbmb" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video", "audio" ], "modalities_out": [ "text", "audio" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-01-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "End-side omni models that take images, video, text and audio and answer in text and speech, for real-time voice and video conversation on devices.", "verified_on": "2026-10-12", "sources": [ "https://github.com/OpenBMB/MiniCPM-o", "https://huggingface.co/api/models?pipeline_tag=any-to-any&sort=downloads&limit=50", "https://www.modelscope.cn/models/OpenBMB/MiniCPM-o-4_5", "https://huggingface.co/openbmb/MiniCPM-o-2_6", "https://huggingface.co/openbmb/MiniCPM-o-4_5" ], "unverified": [] }, { "id": "openbmb-minicpm-o-2-6", "name": "MiniCPM-o 2.6", "level": "variant", "family_id": "openbmb-minicpm-o", "developer": [ "openbmb" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video", "audio" ], "modalities_out": [ "text", "audio" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8674997028, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-01-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: multilingual", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MiniCPM-o 2.6, a checkpoint of the MiniCPM-o family. End-side omni models that take images, video, text and audio and answer in text and speech, for real-time voice and video conversation on devices.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/openbmb/MiniCPM-o-2_6", "https://huggingface.co/api/models?pipeline_tag=any-to-any&sort=downloads&limit=50", "https://www.modelscope.cn/models/OpenBMB/MiniCPM-o-4_5" ], "unverified": [] }, { "id": "openbmb-minicpm-o-4-5", "name": "MiniCPM-o 4.5", "level": "variant", "family_id": "openbmb-minicpm-o", "developer": [ "openbmb" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video", "audio" ], "modalities_out": [ "text", "audio" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 9371787666, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2026-02-03", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MiniCPM-o 4.5, a checkpoint of the MiniCPM-o family. End-side omni models that take images, video, text and audio and answer in text and speech, for real-time voice and video conversation on devices.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/openbmb/MiniCPM-o-4_5", "https://huggingface.co/api/models?pipeline_tag=any-to-any&sort=downloads&limit=50", "https://www.modelscope.cn/models/OpenBMB/MiniCPM-o-4_5" ], "unverified": [] }, { "id": "openbmb-minicpm5-1b", "name": "MiniCPM5 1B", "level": "model", "family_id": null, "developer": [ "openbmb" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 1080632832, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2026-05-21", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "laptop", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, zh", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "One-billion-parameter on-device language model of the MiniCPM5 generation.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/openbmb/MiniCPM5-1B", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://huggingface.co/OpenVINO/MiniCPM5-1B-int4-ov" ], "unverified": [] }, { "id": "deepseek-janus", "name": "Janus", "level": "family", "family_id": null, "developer": [ "deepseek" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text", "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-10-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Unified models that both understand and generate images, using separate visual encoding paths for the two jobs inside one transformer.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2410.13848", "https://github.com/deepseek-ai/Janus", "https://huggingface.co/api/models?pipeline_tag=any-to-any&sort=likes&limit=50", "https://www.kaggle.com/models/deepseek-ai/janus-pro", "https://huggingface.co/deepseek-ai/Janus-1.3B", "https://huggingface.co/deepseek-ai/Janus-Pro-1B" ], "unverified": [] }, { "id": "deepseek-janus-1-3b", "name": "Janus 1.3B", "level": "variant", "family_id": "deepseek-janus", "developer": [ "deepseek" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text", "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 2089297547, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-10-18", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Janus 1.3B, a checkpoint of the Janus family. Unified models that both understand and generate images, using separate visual encoding paths for the two jobs inside one transformer.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2410.13848", "https://huggingface.co/deepseek-ai/Janus-1.3B", "https://huggingface.co/api/models?pipeline_tag=any-to-any&sort=likes&limit=50", "https://www.kaggle.com/models/deepseek-ai/janus-pro" ], "unverified": [] }, { "id": "deepseek-janus-pro-1b", "name": "Janus-Pro 1B", "level": "variant", "family_id": "deepseek-janus", "developer": [ "deepseek" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text", "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-01-29", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Janus-Pro 1B, a checkpoint of the Janus family. Unified models that both understand and generate images, using separate visual encoding paths for the two jobs inside one transformer.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2501.17811", "https://huggingface.co/deepseek-ai/Janus-Pro-1B", "https://huggingface.co/api/models?pipeline_tag=any-to-any&sort=likes&limit=50", "https://www.kaggle.com/models/deepseek-ai/janus-pro" ], "unverified": [] }, { "id": "deepseek-janus-pro-7b", "name": "Janus-Pro 7B", "level": "variant", "family_id": "deepseek-janus", "developer": [ "deepseek" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text", "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-01-29", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Janus-Pro 7B, a checkpoint of the Janus family. Unified models that both understand and generate images, using separate visual encoding paths for the two jobs inside one transformer.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2501.17811", "https://huggingface.co/deepseek-ai/Janus-Pro-7B", "https://huggingface.co/api/models?pipeline_tag=any-to-any&sort=likes&limit=50", "https://www.kaggle.com/models/deepseek-ai/janus-pro" ], "unverified": [] }, { "id": "deepseek-janusflow-1-3b", "name": "JanusFlow 1.3B", "level": "variant", "family_id": "deepseek-janus", "developer": [ "deepseek" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text", "image" ], "architecture": "flow matching", "architecture_detail": null, "params_total": 2046367504, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-11-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "JanusFlow 1.3B, a checkpoint of the Janus family. Unified models that both understand and generate images, using separate visual encoding paths for the two jobs inside one transformer.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2411.07975", "https://huggingface.co/deepseek-ai/JanusFlow-1.3B", "https://huggingface.co/api/models?pipeline_tag=any-to-any&sort=likes&limit=50", "https://www.kaggle.com/models/deepseek-ai/janus-pro" ], "unverified": [] }, { "id": "deepseek-deepseek-math-7b-instruct", "name": "DeepSeekMath 7B Instruct", "level": "model", "family_id": null, "developer": [ "deepseek" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-02-05", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "deepseek-model-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "7B model further pre-trained on 120 billion maths tokens and tuned for step-by-step mathematical reasoning; introduced the GRPO training method.", "verified_on": "2026-10-12", "sources": [ "https://github.com/deepseek-ai/DeepSeek-Math", "https://huggingface.co/deepseek-ai/deepseek-math-7b-instruct", "https://www.kaggle.com/models/deepseek-ai/deepseek-math", "https://www.kaggle.com/models/deepseek-ai/deepseek-prover" ], "unverified": [] }, { "id": "deepseek-deepseek-prover-v2-7b", "name": "DeepSeek-Prover-V2 7B", "level": "model", "family_id": null, "developer": [ "deepseek" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 6910365696, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-04-30", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Model that writes formal proofs in the Lean 4 theorem prover, trained with recursive subgoal decomposition.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/deepseek-ai/DeepSeek-Prover-V2-7B", "https://www.kaggle.com/models/deepseek-ai/deepseek-math", "https://www.kaggle.com/models/deepseek-ai/deepseek-prover" ], "unverified": [] }, { "id": "moonshot-ai-kimi-vl", "name": "Kimi-VL", "level": "family", "family_id": null, "developer": [ "moonshot-ai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-04-09", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Efficient open mixture-of-experts vision language model with about 3 billion active parameters, with instruct and long-thinking versions.", "verified_on": "2026-10-12", "sources": [ "https://www.modelscope.cn/models/moonshotai/Kimi-VL-A3B-Thinking", "https://huggingface.co/moonshotai/Kimi-VL-A3B-Instruct", "https://huggingface.co/moonshotai/Kimi-VL-A3B-Thinking" ], "unverified": [] }, { "id": "moonshot-ai-kimi-vl-a3b-instruct", "name": "Kimi-VL-A3B-Instruct", "level": "variant", "family_id": "moonshot-ai-kimi-vl", "developer": [ "moonshot-ai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 16407657776, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-04-09", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Kimi-VL-A3B-Instruct, a checkpoint of the Kimi-VL family. Efficient open mixture-of-experts vision language model with about 3 billion active parameters, with instruct and long-thinking versions.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/moonshotai/Kimi-VL-A3B-Instruct", "https://www.modelscope.cn/models/moonshotai/Kimi-VL-A3B-Thinking" ], "unverified": [] }, { "id": "moonshot-ai-kimi-vl-a3b-thinking", "name": "Kimi-VL-A3B-Thinking", "level": "variant", "family_id": "moonshot-ai-kimi-vl", "developer": [ "moonshot-ai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 16407657776, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-04-09", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Kimi-VL-A3B-Thinking, a checkpoint of the Kimi-VL family. Efficient open mixture-of-experts vision language model with about 3 billion active parameters, with instruct and long-thinking versions.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/moonshotai/Kimi-VL-A3B-Thinking", "https://www.modelscope.cn/models/moonshotai/Kimi-VL-A3B-Thinking" ], "unverified": [] }, { "id": "stepfun-step3-vl-10b", "name": "Step3-VL-10B", "level": "model", "family_id": null, "developer": [ "stepfun" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 10171750144, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2026-01-13", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "10-billion-parameter open vision language model from StepFun for image understanding and reasoning.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/stepfun-ai/Step3-VL-10B", "https://www.modelscope.cn/models/stepfun-ai/Step3-VL-10B" ], "unverified": [] }, { "id": "stepfun-step-audio", "name": "Step-Audio", "level": "family", "family_id": null, "developer": [ "stepfun" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "audio" ], "modalities_out": [ "text", "audio" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-02-13", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "StepFun speech language models for spoken conversation and audio reasoning, from the 130B Step-Audio-Chat to the R1 reasoning versions.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/api/models?pipeline_tag=audio-text-to-text&sort=likes&limit=50", "https://huggingface.co/stepfun-ai/Step-Audio-Chat", "https://huggingface.co/stepfun-ai/Step-Audio-R1.1" ], "unverified": [] }, { "id": "stepfun-step-audio-chat", "name": "Step-Audio-Chat", "level": "variant", "family_id": "stepfun-step-audio", "developer": [ "stepfun" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "audio" ], "modalities_out": [ "text", "audio" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 131946590208, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-02-13", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Step-Audio-Chat, a checkpoint of the Step-Audio family. StepFun speech language models for spoken conversation and audio reasoning, from the 130B Step-Audio-Chat to the R1 reasoning versions.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/stepfun-ai/Step-Audio-Chat", "https://huggingface.co/api/models?pipeline_tag=audio-text-to-text&sort=likes&limit=50" ], "unverified": [] }, { "id": "stepfun-step-audio-r1-1", "name": "Step-Audio-R1.1", "level": "variant", "family_id": "stepfun-step-audio", "developer": [ "stepfun" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "audio" ], "modalities_out": [ "text", "audio" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 33487033600, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2026-01-14", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Step-Audio-R1.1, a checkpoint of the Step-Audio family. StepFun speech language models for spoken conversation and audio reasoning, from the 130B Step-Audio-Chat to the R1 reasoning versions.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/stepfun-ai/Step-Audio-R1.1", "https://huggingface.co/api/models?pipeline_tag=audio-text-to-text&sort=likes&limit=50" ], "unverified": [] }, { "id": "kyutai-moshiko", "name": "Moshi (Moshiko)", "level": "model", "family_id": null, "developer": [ "kyutai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "audio" ], "modalities_out": [ "text", "audio" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 7687729152, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-09-11", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Full-duplex speech-text foundation model for real-time spoken dialogue, listening and speaking at the same time; Moshiko is the male-voice version.", "verified_on": "2026-10-12", "sources": [ "https://github.com/kyutai-labs/moshi", "https://huggingface.co/kyutai/moshiko-pytorch-bf16", "https://huggingface.co/api/models?pipeline_tag=audio-to-audio&sort=downloads&limit=50" ], "unverified": [] }, { "id": "fixie-ai-ultravox", "name": "Ultravox", "level": "family", "family_id": null, "developer": [ "fixie-ai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "audio" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-02-05", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Checkpoint model cards list 42 languages", "indonesian_support": "not stated", "regional_languages": [ "Thai", "Vietnamese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Speech language models that join a Whisper encoder to an open LLM such as Llama through a trained adapter, so the model takes speech and text input directly.", "verified_on": "2026-10-12", "sources": [ "https://github.com/fixie-ai/ultravox", "https://huggingface.co/api/models?pipeline_tag=audio-text-to-text&sort=downloads&limit=50", "https://huggingface.co/fixie-ai/ultravox-v0_5-llama-3_2-1b", "https://huggingface.co/fixie-ai/ultravox-v0_5-llama-3_1-8b" ], "unverified": [] }, { "id": "fixie-ai-ultravox-v0-5-llama-3-2-1b", "name": "Ultravox v0.5 Llama 3.2 1B", "level": "variant", "family_id": "fixie-ai-ultravox", "developer": [ "fixie-ai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "audio" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 683118592, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-02-06", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 42 languages", "indonesian_support": "not stated", "regional_languages": [ "Thai", "Vietnamese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Ultravox v0.5 Llama 3.2 1B, a checkpoint of the Ultravox family. Speech language models that join a Whisper encoder to an open LLM such as Llama through a trained adapter, so the model takes speech and text input directly.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/fixie-ai/ultravox-v0_5-llama-3_2-1b", "https://huggingface.co/api/models?pipeline_tag=audio-text-to-text&sort=downloads&limit=50" ], "unverified": [] }, { "id": "fixie-ai-ultravox-v0-5-llama-3-1-8b", "name": "Ultravox v0.5 Llama 3.1 8B", "level": "variant", "family_id": "fixie-ai-ultravox", "developer": [ "fixie-ai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "audio" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 687312896, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-02-05", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 42 languages", "indonesian_support": "not stated", "regional_languages": [ "Thai", "Vietnamese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Ultravox v0.5 Llama 3.1 8B, a checkpoint of the Ultravox family. Speech language models that join a Whisper encoder to an open LLM such as Llama through a trained adapter, so the model takes speech and text input directly.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/fixie-ai/ultravox-v0_5-llama-3_1-8b", "https://huggingface.co/api/models?pipeline_tag=audio-text-to-text&sort=downloads&limit=50" ], "unverified": [] }, { "id": "nvidia-audio-flamingo-3", "name": "Audio Flamingo 3", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "audio" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8267215360, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-10-24", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open large audio language model for understanding and reasoning over speech, sound and music; non-commercial licence.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/nvidia/audio-flamingo-3-hf", "https://huggingface.co/api/models?pipeline_tag=audio-text-to-text&sort=downloads&limit=50" ], "unverified": [] }, { "id": "sarvam-ai-shuka-1", "name": "Shuka v1", "level": "model", "family_id": null, "developer": [ "sarvam-ai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "audio" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8725780480, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-08-08", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "llama-3-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, hi", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Audio language model that joins Sarvam's audio encoder to Llama 3 8B so it can answer spoken questions in Indian languages.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/sarvamai/shuka-1", "https://huggingface.co/api/models?pipeline_tag=audio-text-to-text&sort=likes&limit=50" ], "unverified": [] }, { "id": "kuaishou-keye-vl", "name": "Keye-VL", "level": "family", "family_id": null, "developer": [ "kuaishou" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-06-26", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Kuaishou vision language models tuned for short-video understanding as well as general image tasks.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/api/models?pipeline_tag=video-text-to-text&sort=downloads&limit=50", "https://huggingface.co/Kwai-Keye/Keye-VL-8B-Preview", "https://huggingface.co/Kwai-Keye/Keye-VL-1_5-8B" ], "unverified": [] }, { "id": "kuaishou-keye-vl-8b-preview", "name": "Keye-VL 8B Preview", "level": "variant", "family_id": "kuaishou-keye-vl", "developer": [ "kuaishou" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8696828736, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-06-26", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Keye-VL 8B Preview, a checkpoint of the Keye-VL family. Kuaishou vision language models tuned for short-video understanding as well as general image tasks.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Kwai-Keye/Keye-VL-8B-Preview", "https://huggingface.co/api/models?pipeline_tag=video-text-to-text&sort=downloads&limit=50" ], "unverified": [] }, { "id": "kuaishou-keye-vl-1-5-8b", "name": "Keye-VL 1.5 8B", "level": "variant", "family_id": "kuaishou-keye-vl", "developer": [ "kuaishou" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8696835648, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-08-26", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Keye-VL 1.5 8B, a checkpoint of the Keye-VL family. Kuaishou vision language models tuned for short-video understanding as well as general image tasks.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Kwai-Keye/Keye-VL-1_5-8B", "https://huggingface.co/api/models?pipeline_tag=video-text-to-text&sort=downloads&limit=50" ], "unverified": [] }, { "id": "microsoft-fara-7b", "name": "Fara-7B", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8292166656, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-10-30", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small agentic model that operates a computer from screenshots, predicting clicks and typing to complete web tasks.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/microsoft/Fara-7B", "https://ai.azure.com/catalog/models?task=image-to-text" ], "unverified": [] }, { "id": "paige-prism", "name": "Prism", "level": "model", "family_id": null, "developer": [ "paige" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 557695232, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-07-24", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-nd-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Pathology vision language model that writes reports from whole-slide image embeddings produced by Paige's Virchow encoder; research licence.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/paige-ai/Prism", "https://ai.azure.com/catalog/models?industry=health-and-life-sciences" ], "unverified": [] }, { "id": "microsoft-magma-8b", "name": "Magma-8B", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8903066496, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-02-23", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multimodal agentic foundation model that understands images and video and predicts actions for UI navigation and robot manipulation.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/microsoft/Magma-8B", "https://www.modelscope.cn/models/LLM-Research/Magma-8B" ], "unverified": [] }, { "id": "harvard-university-txagent-t1-llama-3-1-8b", "name": "TxAgent T1 Llama 3.1 8B", "level": "model", "family_id": null, "developer": [ "harvard-university" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8030261248, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-03-03", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "llama-3-1-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Agent model from Harvard's Zitnik lab that reasons over biomedical tools to give treatment recommendations, built on Llama 3.1 8B.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/mims-harvard/TxAgent-T1-Llama-3.1-8B", "https://ai.azure.com/catalog/models?industry=health-and-life-sciences" ], "unverified": [] }, { "id": "poolside-laguna-s-2-1", "name": "Laguna S 2.1", "level": "model", "family_id": null, "developer": [ "poolside" ], "model_class": "code", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 117561977600, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2026-07-13", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "openmdw-1-1", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open-weight model from poolside for agentic coding and reasoning.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/poolside/Laguna-S-2.1", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "nvidia-nemotron-nano-9b-v2", "name": "NVIDIA Nemotron Nano 9B v2", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Mamba-2 and transformer hybrid", "params_total": 8888227328, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-08-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "nvidia-open-model-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, es, fr, de, it, ja", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Hybrid Mamba-transformer model for reasoning and chat with a switchable thinking budget, trained from scratch by NVIDIA.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/nvidia/NVIDIA-Nemotron-Nano-9B-v2", "https://docs.aws.amazon.com/bedrock/latest/userguide/model-cards.html" ], "unverified": [] }, { "id": "nvidia-nemotron-nano-12b-v2-vl", "name": "NVIDIA Nemotron Nano 12B v2 VL", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Mamba-2 and transformer hybrid", "params_total": 13181860358, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-10-21", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "nvidia-open-model-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vision language model for document intelligence and video understanding built on the hybrid Nemotron Nano 12B v2 language model.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/nvidia/NVIDIA-Nemotron-Nano-12B-v2-VL-BF16", "https://docs.aws.amazon.com/bedrock/latest/userguide/model-cards.html" ], "unverified": [] }, { "id": "nvidia-llama-3-1-nemotron-nano-8b-v1", "name": "Llama-3.1-Nemotron-Nano-8B-v1", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8030261248, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-03-16", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "nvidia-open-model-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Llama 3.1 8B post-trained by NVIDIA for reasoning with a reasoning on or off switch, sized for a single GPU.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/nvidia/Llama-3.1-Nemotron-Nano-8B-v1", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "nvidia-nemotron-h", "name": "Nemotron-H", "level": "family", "family_id": null, "developer": [ "nvidia" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Mamba-2 layers with a few self-attention layers", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03-19", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, de, es, fr, it, ko, pt, ru, jp, zh", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Hybrid Mamba-transformer base models that replace most attention layers with Mamba-2 layers for faster inference on long sequences.", "verified_on": "2026-10-12", "sources": [ "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/nemo/models/nemotron-h-47b-base-8k", "https://huggingface.co/nvidia/Nemotron-H-8B-Base-8K", "https://huggingface.co/nvidia/Nemotron-H-56B-Base-8K" ], "unverified": [] }, { "id": "nvidia-nemotron-h-8b-base-8k", "name": "Nemotron-H 8B base", "level": "variant", "family_id": "nvidia-nemotron-h", "developer": [ "nvidia" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": null, "params_total": 8100852736, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-03-19", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, de, es, fr, it, ko, pt, ru, jp, zh", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Nemotron-H 8B base, a checkpoint of the Nemotron-H family. Hybrid Mamba-transformer base models that replace most attention layers with Mamba-2 layers for faster inference on long sequences.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/nvidia/Nemotron-H-8B-Base-8K", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/nemo/models/nemotron-h-47b-base-8k" ], "unverified": [] }, { "id": "nvidia-nemotron-h-56b-base-8k", "name": "Nemotron-H 56B base", "level": "variant", "family_id": "nvidia-nemotron-h", "developer": [ "nvidia" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": null, "params_total": 56324350464, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-04-08", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en, de, es, fr, it, ko, pt, ru, jp, zh", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Nemotron-H 56B base, a checkpoint of the Nemotron-H family. Hybrid Mamba-transformer base models that replace most attention layers with Mamba-2 layers for faster inference on long sequences.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/nvidia/Nemotron-H-56B-Base-8K", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/nemo/models/nemotron-h-47b-base-8k" ], "unverified": [] }, { "id": "nvidia-nemotron-mini-4b-instruct", "name": "Nemotron-Mini-4B-Instruct", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-09-10", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU", "laptop" ], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small model pruned and distilled from Nemotron-4 15B, tuned for roleplay, retrieval-augmented question answering and function calling on device.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/nvidia/Nemotron-Mini-4B-Instruct", "https://catalog.ngc.nvidia.com/orgs/nvidia/models/nemotoron-mini-4b-instruct-onnx-int4-rtx" ], "unverified": [] }, { "id": "apple-openelm", "name": "OpenELM", "level": "family", "family_id": null, "developer": [ "apple" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-04-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "apple-amlr", "access_routes": [ "open weights download" ], "deployment_tiers": [ "laptop", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Efficient open language models (270M to 3B) that vary parameters layer by layer, released with the full training and inference framework.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2404.14619", "https://github.com/apple/corenet", "https://huggingface.co/apple/OpenELM", "https://huggingface.co/apple/OpenELM-270M-Instruct", "https://huggingface.co/apple/OpenELM-1_1B-Instruct" ], "unverified": [] }, { "id": "apple-openelm-270m-instruct", "name": "OpenELM 270M Instruct", "level": "variant", "family_id": "apple-openelm", "developer": [ "apple" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 271527168, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-04-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "apple-amlr", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenELM 270M Instruct, a checkpoint of the OpenELM family. Efficient open language models (270M to 3B) that vary parameters layer by layer, released with the full training and inference framework.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2404.14619", "https://huggingface.co/apple/OpenELM-270M-Instruct", "https://huggingface.co/apple/OpenELM" ], "unverified": [] }, { "id": "apple-openelm-1-1b-instruct", "name": "OpenELM 1.1B Instruct", "level": "variant", "family_id": "apple-openelm", "developer": [ "apple" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 1079891456, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-04-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "apple-amlr", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenELM 1.1B Instruct, a checkpoint of the OpenELM family. Efficient open language models (270M to 3B) that vary parameters layer by layer, released with the full training and inference framework.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2404.14619", "https://huggingface.co/apple/OpenELM-1_1B-Instruct", "https://huggingface.co/apple/OpenELM" ], "unverified": [] }, { "id": "apple-openelm-3b-instruct", "name": "OpenELM 3B Instruct", "level": "variant", "family_id": "apple-openelm", "developer": [ "apple" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 3036647424, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-04-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "apple-amlr", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenELM 3B Instruct, a checkpoint of the OpenELM family. Efficient open language models (270M to 3B) that vary parameters layer by layer, released with the full training and inference framework.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2404.14619", "https://huggingface.co/apple/OpenELM-3B-Instruct", "https://huggingface.co/apple/OpenELM" ], "unverified": [] }, { "id": "apple-fastvlm", "name": "FastVLM", "level": "family", "family_id": null, "developer": [ "apple" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "FastViTHD hybrid convolution-transformer vision encoder with an LLM", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-12-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "apple-amlr", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU", "laptop" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Vision language models with a hybrid vision encoder that emits fewer tokens for high-resolution images, cutting time to first token on devices.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2412.13303", "https://github.com/apple/ml-fastvlm", "https://huggingface.co/apple/FastVLM-0.5B", "https://huggingface.co/apple/FastVLM-7B" ], "unverified": [] }, { "id": "apple-fastvlm-0-5b", "name": "FastVLM 0.5B", "level": "variant", "family_id": "apple-fastvlm", "developer": [ "apple" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": null, "params_total": 758833760, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-08-25", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "apple-amlr", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "FastVLM 0.5B, a checkpoint of the FastVLM family. Vision language models with a hybrid vision encoder that emits fewer tokens for high-resolution images, cutting time to first token on devices.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2412.13303", "https://huggingface.co/apple/FastVLM-0.5B" ], "unverified": [] }, { "id": "apple-fastvlm-7b", "name": "FastVLM 7B", "level": "variant", "family_id": "apple-fastvlm", "developer": [ "apple" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": null, "params_total": 7764588000, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-08-25", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "apple-amlr", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "FastVLM 7B, a checkpoint of the FastVLM family. Vision language models with a hybrid vision encoder that emits fewer tokens for high-resolution images, cutting time to first token on devices.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2412.13303", "https://huggingface.co/apple/FastVLM-7B", "https://huggingface.co/apple/FastVLM-0.5B" ], "unverified": [] }, { "id": "apple-dclm-7b", "name": "DCLM-7B", "level": "model", "family_id": null, "developer": [ "apple" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 6889674752, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-07-11", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "apple-ascl", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "7B language model trained on the DCLM-Baseline dataset from the DataComp for Language Models benchmark, released to show the value of data curation.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/apple/DCLM-7B" ], "unverified": [] }, { "id": "meta-mobilellm", "name": "MobileLLM", "level": "family", "family_id": null, "developer": [ "meta" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-02-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Sub-billion-parameter language models with deep-and-thin designs and weight sharing, built for on-device use; non-commercial licence.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2402.14905", "https://github.com/facebookresearch/MobileLLM", "https://www.kaggle.com/models/metaresearch/mobilellm", "https://huggingface.co/facebook/MobileLLM-125M", "https://huggingface.co/facebook/MobileLLM-1B" ], "unverified": [] }, { "id": "meta-mobilellm-125m", "name": "MobileLLM 125M", "level": "variant", "family_id": "meta-mobilellm", "developer": [ "meta" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-10-30", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MobileLLM 125M, a checkpoint of the MobileLLM family. Sub-billion-parameter language models with deep-and-thin designs and weight sharing, built for on-device use; non-commercial licence.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2402.14905", "https://huggingface.co/facebook/MobileLLM-125M", "https://www.kaggle.com/models/metaresearch/mobilellm" ], "unverified": [] }, { "id": "meta-mobilellm-1b", "name": "MobileLLM 1B", "level": "variant", "family_id": "meta-mobilellm", "developer": [ "meta" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-10-31", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MobileLLM 1B, a checkpoint of the MobileLLM family. Sub-billion-parameter language models with deep-and-thin designs and weight sharing, built for on-device use; non-commercial licence.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2402.14905", "https://huggingface.co/facebook/MobileLLM-1B", "https://www.kaggle.com/models/metaresearch/mobilellm" ], "unverified": [] }, { "id": "open-thoughts-openthinker-7b", "name": "OpenThinker-7B", "level": "model", "family_id": null, "developer": [ "open-thoughts" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 7615616512, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-01-28", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen2.5-7B fine-tuned on the open OpenThoughts reasoning dataset, released with its data to reproduce reasoning models.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/open-thoughts/OpenThinker-7B", "https://huggingface.co/api/models?pipeline_tag=reinforcement-learning&sort=downloads&limit=50", "https://www.kaggle.com/models/open-thoughts/openthinker" ], "unverified": [] }, { "id": "ai21-labs-jamba-reasoning-3b", "name": "Jamba Reasoning 3B", "level": "model", "family_id": null, "developer": [ "ai21-labs" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "hybrid", "architecture_detail": "Transformer and Mamba hybrid", "params_total": 3197109632, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-10-05", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "laptop", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small hybrid transformer-Mamba reasoning model with a long context window, sized to run on laptops and phones.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/ai21labs/AI21-Jamba-Reasoning-3B", "https://www.kaggle.com/models/ai21labs/ai21-jamba-reasoning-3b" ], "unverified": [] }, { "id": "alibaba-qwen3-235b-a22b", "name": "Qwen3-235B-A22B", "level": "model", "family_id": "alibaba-qwen3", "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 235093634560, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-04-27", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Flagship open mixture-of-experts model of the Qwen3 generation, 235B total and 22B active parameters, with switchable thinking and non-thinking modes.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Qwen/Qwen3-235B-A22B", "https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/maas/use-open-models" ], "unverified": [] }, { "id": "alibaba-qwen3-32b", "name": "Qwen3-32B", "level": "model", "family_id": "alibaba-qwen3", "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 32762123264, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-04-27", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Largest dense model of the Qwen3 generation, with switchable thinking and non-thinking modes.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Qwen/Qwen3-32B", "https://docs.aws.amazon.com/bedrock/latest/userguide/model-cards.html" ], "unverified": [] }, { "id": "alibaba-qwen3-30b-a3b", "name": "Qwen3-30B-A3B", "level": "model", "family_id": "alibaba-qwen3", "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 30532122624, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-04-27", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small mixture-of-experts Qwen3 model with 30B total and 3B active parameters, with switchable thinking and non-thinking modes.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Qwen/Qwen3-30B-A3B", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "alibaba-qwen3-next-80b-a3b-instruct", "name": "Qwen3-Next-80B-A3B-Instruct", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": "Hybrid Gated DeltaNet and gated attention with high-sparsity MoE", "params_total": 81324862720, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-09-09", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen3-Next model with hybrid linear and full attention and an ultra-sparse MoE (80B total, about 3B active) for fast long-context inference.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Qwen/Qwen3-Next-80B-A3B-Instruct", "https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/maas/use-open-models", "https://www.modelscope.cn/models/Qwen/Qwen3-Next-80B-A3B-Instruct" ], "unverified": [] }, { "id": "alibaba-qwen3-vl-235b-a22b-instruct", "name": "Qwen3-VL-235B-A22B-Instruct", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 235670022896, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-09-22", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Largest Qwen3-VL vision language model, for image and video understanding, visual agents and long documents.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Qwen/Qwen3-VL-235B-A22B-Instruct", "https://docs.aws.amazon.com/bedrock/latest/userguide/model-cards.html" ], "unverified": [] }, { "id": "alibaba-qwen3-omni-30b-a3b-instruct", "name": "Qwen3-Omni-30B-A3B-Instruct", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video", "audio" ], "modalities_out": [ "text", "audio" ], "architecture": "MoE", "architecture_detail": null, "params_total": 35259818545, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-09-20", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Natively omni-modal Qwen3 model that takes text, images, audio and video and answers in text and real-time speech.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Qwen/Qwen3-Omni-30B-A3B-Instruct", "https://huggingface.co/api/models?pipeline_tag=any-to-any&sort=downloads&limit=50", "https://www.modelscope.cn/models/Qwen/Qwen3-Omni-30B-A3B-Instruct" ], "unverified": [] }, { "id": "alibaba-qwen2-5-omni-7b", "name": "Qwen2.5-Omni-7B", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image", "video", "audio" ], "modalities_out": [ "text", "audio" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 10732225440, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-03-22", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "End-to-end multimodal model (Thinker-Talker design) that perceives text, images, audio and video and answers in text and streaming speech.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Qwen/Qwen2.5-Omni-7B", "https://huggingface.co/api/models?pipeline_tag=any-to-any&sort=downloads&limit=50", "https://www.modelscope.cn/models/Qwen/Qwen2.5-Omni-7B" ], "unverified": [] }, { "id": "alibaba-qwen2-audio-7b-instruct", "name": "Qwen2-Audio-7B-Instruct", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "audio" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8397094912, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-07-31", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Audio language model that accepts speech and sound and answers in text, for voice chat and audio analysis.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Qwen/Qwen2-Audio-7B-Instruct", "https://huggingface.co/api/models?pipeline_tag=audio-text-to-text&sort=downloads&limit=50", "https://www.kaggle.com/models/qwen-lm/qwen2-audio" ], "unverified": [] }, { "id": "alibaba-qwen-vl-chat", "name": "Qwen-VL-Chat", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-08-20", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: zh, en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First-generation Qwen vision language chat model, for image understanding, grounding and text reading in Chinese and English.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Qwen/Qwen-VL-Chat", "https://replicate.com/lucataco/qwen-vl-chat" ], "unverified": [] }, { "id": "alibaba-qvq-72b-preview", "name": "QVQ-72B-Preview", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "reasoning", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 73405560320, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-12-24", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": "Experimental research model focused on visual reasoning, built on Qwen2-VL-72B.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Qwen/QVQ-72B-Preview", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://www.kaggle.com/models/qwen-lm/qvq-72b-preview" ], "unverified": [] }, { "id": "alibaba-qwen-agentworld-35b-a3b", "name": "Qwen-AgentWorld-35B-A3B", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 34660610688, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2026-06-23", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Language world model from Qwen that simulates environments for general agents, a mixture-of-experts model (35B total, 3B active parameters by its name).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2606.24597", "https://huggingface.co/Qwen/Qwen-AgentWorld-35B-A3B", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://keras.io/keras_hub/presets/" ], "unverified": [] }, { "id": "alibaba-qwen3-vl-reranker", "name": "Qwen3-VL-Reranker", "level": "family", "family_id": null, "developer": [ "alibaba" ], "model_class": "reranker", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01-07", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multimodal rerankers built on Qwen3-VL that score text, image and document candidates against a query; companions to the Qwen3-VL-Embedding records.", "verified_on": "2026-10-12", "sources": [ "https://www.modelscope.cn/models/Qwen/Qwen3-VL-Reranker-8B", "https://huggingface.co/Qwen/Qwen3-VL-Reranker-2B", "https://huggingface.co/Qwen/Qwen3-VL-Reranker-8B" ], "unverified": [] }, { "id": "alibaba-qwen3-vl-reranker-2b", "name": "Qwen3-VL-Reranker-2B", "level": "variant", "family_id": "alibaba-qwen3-vl-reranker", "developer": [ "alibaba" ], "model_class": "reranker", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 2127532032, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2026-01-07", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen3-VL-Reranker-2B, a checkpoint of the Qwen3-VL-Reranker family. Multimodal rerankers built on Qwen3-VL that score text, image and document candidates against a query; companions to the Qwen3-VL-Embedding records.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Qwen/Qwen3-VL-Reranker-2B", "https://www.modelscope.cn/models/Qwen/Qwen3-VL-Reranker-8B" ], "unverified": [] }, { "id": "alibaba-qwen3-vl-reranker-8b", "name": "Qwen3-VL-Reranker-8B", "level": "variant", "family_id": "alibaba-qwen3-vl-reranker", "developer": [ "alibaba" ], "model_class": "reranker", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8767123696, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2026-01-07", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Qwen3-VL-Reranker-8B, a checkpoint of the Qwen3-VL-Reranker family. Multimodal rerankers built on Qwen3-VL that score text, image and document candidates against a query; companions to the Qwen3-VL-Embedding records.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Qwen/Qwen3-VL-Reranker-8B", "https://www.modelscope.cn/models/Qwen/Qwen3-VL-Reranker-8B" ], "unverified": [] }, { "id": "cohere-labs-aya-101", "name": "Aya 101", "level": "model", "family_id": null, "developer": [ "cohere-labs" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 12921057280, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-02-08", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 101 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Javanese", "Sundanese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Massively multilingual instruction-tuned mT5 model that follows instructions in 101 languages, including Indonesian, Javanese and Sundanese.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/CohereLabs/aya-101", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html", "https://www.kaggle.com/models/cohereforai/aya" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "cohere-labs-aya-vision", "name": "Aya Vision", "level": "family", "family_id": null, "developer": [ "cohere-labs" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03-02", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Checkpoint model cards list 23 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Vietnamese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Open-weight multilingual vision language models from Cohere Labs covering 23 languages, including Indonesian; non-commercial licence.", "verified_on": "2026-10-12", "sources": [ "https://www.kaggle.com/models/cohereforai/aya-vision", "https://huggingface.co/CohereLabs/aya-vision-8b", "https://huggingface.co/CohereLabs/aya-vision-32b" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "cohere-labs-aya-vision-8b", "name": "Aya Vision 8B", "level": "variant", "family_id": "cohere-labs-aya-vision", "developer": [ "cohere-labs" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8631842032, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-03-02", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 23 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Vietnamese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Aya Vision 8B, a checkpoint of the Aya Vision family. Open-weight multilingual vision language models from Cohere Labs covering 23 languages, including Indonesian; non-commercial licence.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/CohereLabs/aya-vision-8b", "https://www.kaggle.com/models/cohereforai/aya-vision" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "cohere-labs-aya-vision-32b", "name": "Aya Vision 32B", "level": "variant", "family_id": "cohere-labs-aya-vision", "developer": [ "cohere-labs" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 33137896688, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-03-02", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card lists 23 languages, including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [ "Vietnamese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "Aya Vision 32B, a checkpoint of the Aya Vision family. Open-weight multilingual vision language models from Cohere Labs covering 23 languages, including Indonesian; non-commercial licence.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/CohereLabs/aya-vision-32b", "https://www.kaggle.com/models/cohereforai/aya-vision" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "vietai-vit5-base", "name": "ViT5 base", "level": "model", "family_id": null, "developer": [ "vietai" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-03-14", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Vietnamese", "indonesian_support": "not stated", "regional_languages": [ "Vietnamese" ], "indonesia_hosting": "unknown", "status": "active", "summary": "T5-style encoder-decoder pre-trained on Vietnamese text, for Vietnamese summarisation and generation.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/VietAI/vit5-base", "https://huggingface.co/api/models?pipeline_tag=question-answering&sort=downloads&limit=50" ], "unverified": [] }, { "id": "cahya-t5-base-indonesian-summarization-cased", "name": "T5 base Indonesian summarization", "level": "model", "family_id": null, "developer": [ "cahya" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Indonesian", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "T5 base model fine-tuned to summarise Indonesian news articles (id_liputan6 dataset).", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/cahya/t5-base-indonesian-summarization-cased", "https://huggingface.co/api/models?pipeline_tag=summarization&sort=downloads&limit=50" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "speakleash-bielik-11b-v3-0-instruct", "name": "Bielik 11B v3.0 Instruct", "level": "model", "family_id": null, "developer": [ "speakleash" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 11168796672, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-11-07", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Polish and other European languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Polish-focused 11B instruction model from the SpeakLeash open science project and ACK Cyfronet AGH.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/speakleash/Bielik-11B-v3.0-Instruct", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [ "summary" ] }, { "id": "institute-of-science-tokyo-swallow-7b-instruct", "name": "Swallow 7B Instruct", "level": "model", "family_id": null, "developer": [ "institute-of-science-tokyo" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 6829969408, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-12-07", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "llama-2-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Japanese and English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Llama 2 7B continually pre-trained on Japanese text with an extended Japanese vocabulary, then instruction tuned.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/tokyotech-llm/Swallow-7b-instruct-hf", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "cyberagent-calm3-22b-chat", "name": "CALM3 22B Chat", "level": "model", "family_id": null, "developer": [ "cyberagent" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 22542882816, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-07-01", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Japanese and English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Japanese and English chat model trained from scratch by CyberAgent on 2 trillion tokens.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/cyberagent/calm3-22b-chat", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html" ], "unverified": [] }, { "id": "elyza-llama-3-elyza-jp-8b", "name": "Llama-3-ELYZA-JP-8B", "level": "model", "family_id": null, "developer": [ "elyza" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8030261248, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-06-25", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "llama-3-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Japanese and English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Meta Llama 3 8B further pre-trained and instruction tuned by ELYZA for Japanese.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/elyza/Llama-3-ELYZA-JP-8B", "https://aihub.qualcomm.com/models/llama_v3_elyza_jp_8b" ], "unverified": [] }, { "id": "karakuri-karakuri-lm-8x7b-instruct-v0-1", "name": "KARAKURI LM 8x7B Instruct v0.1", "level": "model", "family_id": null, "developer": [ "karakuri" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 46703841280, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-06-17", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Japanese and English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Japanese mixture-of-experts chat model fine-tuned from Swallow-MX 8x7B for Japanese and English tasks including function calling.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/karakuri-ai/karakuri-lm-8x7b-instruct-v0.1", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "stockmark-stockmark-13b", "name": "Stockmark-13b", "level": "model", "family_id": null, "developer": [ "stockmark" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 13200184320, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-10-21", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Japanese", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "13-billion-parameter language model pre-trained from scratch on a Japanese corpus of about 220 billion tokens.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/stockmark/stockmark-13b", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "kakao-kanana-2-30b-a3b-instruct", "name": "Kanana-2 30B A3B Instruct", "level": "model", "family_id": null, "developer": [ "kakao" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "MoE", "architecture_detail": null, "params_total": 30670809088, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2026-01-14", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Korean and English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Kakao mixture-of-experts model (30B total, 3B active) for Korean and English, tuned for instruction following and agentic tool use.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/kakaocorp/kanana-2-30b-a3b-instruct-2601", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "sdaia-allam-7b-instruct-preview", "name": "ALLaM-7B-Instruct-preview", "level": "model", "family_id": null, "developer": [ "sdaia" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 7000559616, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2025-02-13", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Arabic and English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": "Arabic and English language model trained from scratch by the Saudi Data and AI Authority (SDAIA) to advance Arabic language technology.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/humain-ai/ALLaM-7B-Instruct-preview", "https://huggingface.co/qualcomm/Allam-7B" ], "unverified": [] }, { "id": "taide-llama3-taide-lx-8b-chat-alpha1", "name": "Llama3-TAIDE-LX-8B-Chat-Alpha1", "level": "model", "family_id": null, "developer": [ "taide" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8030261248, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-04-26", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Traditional Chinese and English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Llama 3 8B continually pre-trained and tuned by Taiwan's TAIDE project for Traditional Chinese and Taiwanese culture.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/taide/Llama3-TAIDE-LX-8B-Chat-Alpha1", "https://aihub.qualcomm.com/models/llama_v3_taide_8b_chat" ], "unverified": [ "summary" ] }, { "id": "unitary-unbiased-toxic-roberta", "name": "Detoxify unbiased-toxic-roberta", "level": "model", "family_id": null, "developer": [ "unitary" ], "model_class": "guard", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "RoBERTa classifier from the Detoxify project that flags toxic comments while reducing bias against identity mentions, for content moderation.", "verified_on": "2026-10-12", "sources": [ "https://github.com/unitaryai/detoxify", "https://huggingface.co/unitary/unbiased-toxic-roberta", "https://huggingface.co/api/models?pipeline_tag=text-classification&sort=downloads&limit=50" ], "unverified": [] }, { "id": "vectara-hallucination-evaluation-model", "name": "HHEM-2.1-Open", "level": "model", "family_id": null, "developer": [ "vectara" ], "model_class": "guard", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 109630082, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2023-10-25", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Classifier that scores whether a generated text is supported by its source text, used to detect hallucinations in retrieval-augmented generation.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/vectara/hallucination_evaluation_model", "https://www.kaggle.com/models/vectara/hallucination_evaluation_model" ], "unverified": [] }, { "id": "bespoke-labs-bespoke-minicheck-7b", "name": "Bespoke-MiniCheck-7B", "level": "model", "family_id": null, "developer": [ "bespoke-labs" ], "model_class": "guard", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 7737757696, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-08-09", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: en", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fact-checking model that decides whether a claim is supported by a grounding document, for checking LLM outputs; non-commercial licence.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/bespokelabs/Bespoke-MiniCheck-7B", "https://www.kaggle.com/models/bespoke-labs/minicheck" ], "unverified": [] }, { "id": "bigcode-starpii", "name": "StarPII", "level": "model", "family_id": null, "developer": [ "bigcode" ], "model_class": "guard", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-04-23", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Model card languages: code", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Named entity recognition model that detects personal data (names, emails, keys, passwords, IP addresses) in code, used to clean training data.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/bigcode/starpii", "https://huggingface.co/api/models?pipeline_tag=token-classification&sort=likes&limit=50" ], "unverified": [ "summary" ] }, { "id": "roblox-voice-safety-classifier", "name": "Roblox Voice Safety Classifier", "level": "model", "family_id": null, "developer": [ "roblox" ], "model_class": "guard", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 94580342, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the safetensors index on Hugging Face.", "context_length": null, "release_date": "2024-06-28", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-sa-3-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Audio classifier fine-tuned from WavLM that flags policy violations such as profanity and bullying in voice chat.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Roblox/voice-safety-classifier", "https://huggingface.co/api/models?pipeline_tag=audio-classification&sort=likes&limit=50" ], "unverified": [] }, { "id": "openai-text-embedding", "name": "OpenAI text embeddings", "level": "family", "family_id": null, "developer": [ "openai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenAI API embedding models that turn text into vectors for search, clustering, recommendations and classification; text-embedding-3 small and large replaced ada-002.", "verified_on": "2026-10-12", "sources": [ "https://platform.openai.com/docs/guides/embeddings", "https://learn.microsoft.com/en-us/azure/foundry/foundry-models/concepts/models-sold-directly-by-azure" ], "unverified": [] }, { "id": "openai-text-embedding-3-small", "name": "text-embedding-3-small", "level": "variant", "family_id": "openai-text-embedding", "developer": [ "openai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 8192, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Smaller, lower-cost text-embedding-3 model from OpenAI, offered through the OpenAI API and Azure.", "verified_on": "2026-10-12", "sources": [ "https://platform.openai.com/docs/guides/embeddings", "https://learn.microsoft.com/en-us/azure/foundry/foundry-models/concepts/models-sold-directly-by-azure" ], "unverified": [] }, { "id": "openai-text-embedding-3-large", "name": "text-embedding-3-large", "level": "variant", "family_id": "openai-text-embedding", "developer": [ "openai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 8192, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Larger, most capable text-embedding-3 model from OpenAI; its vectors can be shortened (for example to 256) with little loss.", "verified_on": "2026-10-12", "sources": [ "https://platform.openai.com/docs/guides/embeddings", "https://learn.microsoft.com/en-us/azure/foundry/foundry-models/concepts/models-sold-directly-by-azure" ], "unverified": [] }, { "id": "openai-text-embedding-ada-002", "name": "text-embedding-ada-002", "level": "variant", "family_id": "openai-text-embedding", "developer": [ "openai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 8192, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Earlier OpenAI text embedding model with 1,536-dimension vectors, still listed on Azure; text-embedding-3 models outperform it.", "verified_on": "2026-10-12", "sources": [ "https://platform.openai.com/docs/guides/embeddings", "https://learn.microsoft.com/en-us/azure/foundry/foundry-models/concepts/models-sold-directly-by-azure" ], "unverified": [] }, { "id": "cohere-embed", "name": "Cohere Embed", "level": "family", "family_id": null, "developer": [ "cohere" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Cohere API embedding models for semantic search, retrieval and classification over text and images, in English-only and multilingual versions.", "verified_on": "2026-10-12", "sources": [ "https://docs.cohere.com/docs/cohere-embed", "https://docs.cohere.com/docs/models", "https://docs.aws.amazon.com/bedrock/latest/userguide/model-cards.html" ], "unverified": [] }, { "id": "cohere-embed-english-v3-0", "name": "Embed English v3.0", "level": "variant", "family_id": "cohere-embed", "developer": [ "cohere" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 512, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "English-only Cohere embedding model producing 1,024-dimension vectors from text or images.", "verified_on": "2026-10-12", "sources": [ "https://docs.cohere.com/docs/cohere-embed", "https://learn.microsoft.com/en-us/azure/foundry/foundry-models/concepts/models-sold-directly-by-azure" ], "unverified": [] }, { "id": "cohere-embed-multilingual-v3-0", "name": "Embed Multilingual v3.0", "level": "variant", "family_id": "cohere-embed", "developer": [ "cohere" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 512, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Multilingual; the developer's supported language list includes Indonesian", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multilingual Cohere embedding model producing 1,024-dimension vectors; its supported language list includes Indonesian.", "verified_on": "2026-10-12", "sources": [ "https://docs.cohere.com/docs/cohere-embed", "https://docs.aws.amazon.com/bedrock/latest/userguide/model-cards.html" ], "unverified": [ "indonesian_measured" ] }, { "id": "cohere-embed-v4-0", "name": "Embed v4.0", "level": "variant", "family_id": "cohere-embed", "developer": [ "cohere" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 128000, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Cohere embedding model for text, images and mixed documents such as PDFs, with selectable vector sizes from 256 to 1,536.", "verified_on": "2026-10-12", "sources": [ "https://docs.cohere.com/docs/cohere-embed", "https://docs.aws.amazon.com/bedrock/latest/userguide/model-cards.html" ], "unverified": [] }, { "id": "cohere-embed-v5-0-pro", "name": "Embed v5.0 Pro", "level": "variant", "family_id": "cohere-embed", "developer": [ "cohere" ], "model_class": "embedding", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Cohere's most capable embedding model for text and images; embed-v5.0-fast is the lighter companion.", "verified_on": "2026-10-12", "sources": [ "https://docs.cohere.com/docs/cohere-embed", "https://ai.azure.com/catalog/models?task=embeddings" ], "unverified": [] }, { "id": "cohere-rerank", "name": "Cohere Rerank", "level": "family", "family_id": null, "developer": [ "cohere" ], "model_class": "reranker", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Cohere API rerankers that re-order documents and semi-structured JSON by relevance to a query, in English and other languages.", "verified_on": "2026-10-12", "sources": [ "https://docs.cohere.com/docs/rerank", "https://docs.aws.amazon.com/bedrock/latest/userguide/model-cards.html" ], "unverified": [] }, { "id": "cohere-rerank-v3-5", "name": "Rerank v3.5", "level": "variant", "family_id": "cohere-rerank", "developer": [ "cohere" ], "model_class": "reranker", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Cohere reranker for documents and semi-structured data that supports the same languages as Embed Multilingual v3.0.", "verified_on": "2026-10-12", "sources": [ "https://docs.cohere.com/docs/rerank", "https://docs.aws.amazon.com/bedrock/latest/userguide/model-cards.html" ], "unverified": [ "indonesian_measured" ] }, { "id": "cohere-rerank-v4-0-pro", "name": "Rerank v4.0 Pro", "level": "variant", "family_id": "cohere-rerank", "developer": [ "cohere" ], "model_class": "reranker", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multilingual Cohere reranker for English and non-English documents and JSON, aimed at the highest quality and complex cases.", "verified_on": "2026-10-12", "sources": [ "https://docs.cohere.com/docs/rerank", "https://learn.microsoft.com/en-us/azure/foundry/foundry-models/concepts/models-sold-directly-by-azure" ], "unverified": [] }, { "id": "cohere-rerank-v4-0-fast", "name": "Rerank v4.0 Fast", "level": "variant", "family_id": "cohere-rerank", "developer": [ "cohere" ], "model_class": "reranker", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Lighter version of Rerank v4.0 Pro for low-latency, high-throughput re-ranking.", "verified_on": "2026-10-12", "sources": [ "https://docs.cohere.com/docs/rerank", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "voyage-ai-voyage-embeddings", "name": "Voyage embeddings", "level": "family", "family_id": null, "developer": [ "voyage-ai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Voyage AI API text embedding models for general, multilingual, code and domain retrieval, with 32,000-token inputs for current versions.", "verified_on": "2026-10-12", "sources": [ "https://docs.voyageai.com/docs/embeddings", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "voyage-ai-voyage-3-large", "name": "voyage-3-large", "level": "variant", "family_id": "voyage-ai-voyage-embeddings", "developer": [ "voyage-ai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 32000, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Previous-generation Voyage general-purpose and multilingual text embedding model with selectable vector sizes.", "verified_on": "2026-10-12", "sources": [ "https://docs.voyageai.com/docs/embeddings", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "voyage-ai-voyage-3-5", "name": "voyage-3.5", "level": "variant", "family_id": "voyage-ai-voyage-embeddings", "developer": [ "voyage-ai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 32000, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Previous-generation Voyage text embedding model for general-purpose and multilingual retrieval.", "verified_on": "2026-10-12", "sources": [ "https://docs.voyageai.com/docs/embeddings", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "voyage-ai-voyage-4", "name": "voyage-4", "level": "variant", "family_id": "voyage-ai-voyage-embeddings", "developer": [ "voyage-ai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 32000, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Current-generation Voyage text embedding model for general-purpose and multilingual retrieval; embeddings are compatible across the 4 series.", "verified_on": "2026-10-12", "sources": [ "https://docs.voyageai.com/docs/embeddings", "https://ai.azure.com/catalog/models?task=embeddings" ], "unverified": [] }, { "id": "voyage-ai-voyage-code-3", "name": "voyage-code-3", "level": "variant", "family_id": "voyage-ai-voyage-embeddings", "developer": [ "voyage-ai" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 32000, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Voyage embedding model optimised for code retrieval.", "verified_on": "2026-10-12", "sources": [ "https://docs.voyageai.com/docs/embeddings", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "voyage-ai-voyage-rerank", "name": "Voyage rerankers", "level": "family", "family_id": null, "developer": [ "voyage-ai" ], "model_class": "reranker", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Voyage AI API rerankers that score query and document pairs, from the first-generation rerank-1 and rerank-lite-1 to rerank-3.", "verified_on": "2026-10-12", "sources": [ "https://docs.voyageai.com/docs/reranker", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "voyage-ai-rerank-2-5", "name": "rerank-2.5", "level": "variant", "family_id": "voyage-ai-voyage-rerank", "developer": [ "voyage-ai" ], "model_class": "reranker", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 32000, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Voyage generalist reranker with instruction following and multilingual support.", "verified_on": "2026-10-12", "sources": [ "https://docs.voyageai.com/docs/reranker" ], "unverified": [] }, { "id": "amazon-titan-text-embeddings-v2", "name": "Amazon Titan Text Embeddings V2", "level": "model", "family_id": null, "developer": [ "amazon" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 8192, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English (100+ languages in preview)", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Amazon Bedrock text embedding model that takes up to 8,192 tokens and returns 1,024, 512 or 256-dimension vectors, for retrieval, search and classification.", "verified_on": "2026-10-12", "sources": [ "https://docs.aws.amazon.com/bedrock/latest/userguide/titan-embedding-models.html", "https://docs.aws.amazon.com/bedrock/latest/userguide/model-cards.html" ], "unverified": [] }, { "id": "google-gemini-embedding", "name": "Gemini Embedding", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Embedding models built from Gemini for retrieval, classification and clustering via the Gemini API; gemini-embedding-001 is text, Gemini Embedding 2 maps text, images, video, audio and documents into one space.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2503.07891", "https://ai.google.dev/gemini-api/docs/embeddings", "https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/google-models" ], "unverified": [] }, { "id": "google-gemini-embedding-001", "name": "gemini-embedding-001", "level": "variant", "family_id": "google-gemini-embedding", "developer": [ "google" ], "model_class": "embedding", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Text embedding model from the Gemini Embedding paper, offered through the Gemini API with 3,072-dimension vectors by default.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2503.07891", "https://ai.google.dev/gemini-api/docs/embeddings" ], "unverified": [] }, { "id": "google-gemini-embedding-2", "name": "Gemini Embedding 2", "level": "variant", "family_id": "google-gemini-embedding", "developer": [ "google" ], "model_class": "embedding", "modalities_in": [ "text", "image", "video", "audio" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First multimodal embedding model in the Gemini API, mapping text, images, video, audio and documents into one embedding space for cross-modal search.", "verified_on": "2026-10-12", "sources": [ "https://ai.google.dev/gemini-api/docs/embeddings" ], "unverified": [] }, { "id": "twelvelabs-marengo", "name": "TwelveLabs Marengo", "level": "family", "family_id": null, "developer": [ "twelvelabs" ], "model_class": "embedding", "modalities_in": [ "text", "image", "video", "audio" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Video embedding models that combine visual, audio and text signals for video search and retrieval; versions 3.0 and 3.5 are current.", "verified_on": "2026-10-12", "sources": [ "https://docs.twelvelabs.io/docs/concepts/models/marengo", "https://docs.aws.amazon.com/bedrock/latest/userguide/model-cards.html" ], "unverified": [] }, { "id": "twelvelabs-marengo-3-0", "name": "Marengo 3.0", "level": "variant", "family_id": "twelvelabs-marengo", "developer": [ "twelvelabs" ], "model_class": "embedding", "modalities_in": [ "text", "image", "video", "audio" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "TwelveLabs embedding model for searching content and creating embeddings from video, audio, images and text.", "verified_on": "2026-10-12", "sources": [ "https://docs.twelvelabs.io/docs/concepts/models/marengo", "https://docs.aws.amazon.com/bedrock/latest/userguide/model-cards.html" ], "unverified": [] }, { "id": "twelvelabs-pegasus", "name": "TwelveLabs Pegasus", "level": "family", "family_id": null, "developer": [ "twelvelabs" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "video", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Video language models that analyse videos and generate summaries, descriptions and structured answers.", "verified_on": "2026-10-12", "sources": [ "https://docs.twelvelabs.io/docs/concepts/models/pegasus", "https://docs.aws.amazon.com/bedrock/latest/userguide/model-cards.html" ], "unverified": [] }, { "id": "twelvelabs-pegasus-1-5", "name": "Pegasus 1.5", "level": "variant", "family_id": "twelvelabs-pegasus", "developer": [ "twelvelabs" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "video" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "TwelveLabs model that analyses videos to generate summaries, detailed descriptions and structured responses.", "verified_on": "2026-10-12", "sources": [ "https://docs.twelvelabs.io/docs/concepts/models/pegasus", "https://docs.aws.amazon.com/bedrock/latest/userguide/model-cards.html" ], "unverified": [] }, { "id": "openai-gpt-5-nano", "name": "GPT-5 nano", "level": "variant", "family_id": "openai-gpt-5", "developer": [ "openai" ], "model_class": "reasoning", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 400000, "release_date": "2025-08-07", "release_date_basis": "inferred from page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fastest and cheapest version of GPT-5, suited to summarisation and classification, with a 400,000-token context window.", "verified_on": "2026-10-12", "sources": [ "https://platform.openai.com/docs/models/gpt-5-nano", "https://replicate.com/openai/gpt-5-nano" ], "unverified": [] }, { "id": "openai-gpt-5-pro", "name": "GPT-5 pro", "level": "variant", "family_id": "openai-gpt-5", "developer": [ "openai" ], "model_class": "reasoning", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 400000, "release_date": "2025-10-06", "release_date_basis": "inferred from page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Version of GPT-5 that uses more compute to think harder on tough problems; available only in the Responses API and only at high reasoning effort.", "verified_on": "2026-10-12", "sources": [ "https://platform.openai.com/docs/models/gpt-5-pro", "https://replicate.com/openai/gpt-5-nano" ], "unverified": [] }, { "id": "openai-gpt-5-2-codex", "name": "GPT-5.2-Codex", "level": "variant", "family_id": "openai-gpt-5", "developer": [ "openai" ], "model_class": "code", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 400000, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Upgraded version of GPT-5.2 optimised for agentic coding tasks in Codex or similar environments, with low to xhigh reasoning effort settings.", "verified_on": "2026-10-12", "sources": [ "https://platform.openai.com/docs/models/gpt-5.2-codex", "https://learn.microsoft.com/en-us/azure/foundry/foundry-models/concepts/models-sold-directly-by-azure" ], "unverified": [] }, { "id": "openai-o1-mini", "name": "o1-mini", "level": "variant", "family_id": "openai-o-series", "developer": [ "openai" ], "model_class": "reasoning", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 128000, "release_date": "2024-09-12", "release_date_basis": "inferred from page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Faster, more affordable o1 reasoning model; OpenAI now recommends o3-mini in its place.", "verified_on": "2026-10-12", "sources": [ "https://platform.openai.com/docs/models/o1-mini", "https://replicate.com/openai/o1-mini" ], "unverified": [] }, { "id": "openai-codex-mini-latest", "name": "codex-mini-latest", "level": "model", "family_id": null, "developer": [ "openai" ], "model_class": "code", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 200000, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Fine-tuned version of o4-mini made for use in the Codex CLI, with a 200,000-token context window.", "verified_on": "2026-10-12", "sources": [ "https://platform.openai.com/docs/models/codex-mini-latest", "https://learn.microsoft.com/en-us/azure/foundry/foundry-models/concepts/models-sold-directly-by-azure" ], "unverified": [] }, { "id": "openai-computer-use-preview", "name": "computer-use-preview", "level": "model", "family_id": null, "developer": [ "openai" ], "model_class": "multimodal LLM", "modalities_in": [ "text", "image" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 8192, "release_date": "2025-03-11", "release_date_basis": "inferred from page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": "Specialised model behind the OpenAI computer use tool, trained to understand and carry out computer tasks; usable only in the Responses API.", "verified_on": "2026-10-12", "sources": [ "https://platform.openai.com/docs/models/computer-use-preview", "https://learn.microsoft.com/en-us/azure/foundry/foundry-models/concepts/models-sold-directly-by-azure" ], "unverified": [] }, { "id": "google-deepmind-gemini-3-5-flash-lite", "name": "Gemini 3.5 Flash-Lite", "level": "variant", "family_id": "google-deepmind-gemini", "developer": [ "google-deepmind" ], "model_class": "multimodal LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fast, budget-friendly Gemini model optimised for low latency and high-volume traffic, with tool orchestration and document understanding.", "verified_on": "2026-10-12", "sources": [ "https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/google-models" ], "unverified": [] }, { "id": "google-deepmind-gemini-3-8-flash-cyber", "name": "Gemini 3.8 Flash Cyber", "level": "variant", "family_id": "google-deepmind-gemini", "developer": [ "google-deepmind" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Gemini model optimised for cyber security, listed as generally available on Google Cloud.", "verified_on": "2026-10-12", "sources": [ "https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/google-models" ], "unverified": [] }, { "id": "mistral-ai-ministral-3b-2410", "name": "Ministral 3B (24.10)", "level": "model", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "edge LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "shut down", "summary": "Small Mistral model of October 2024 (ministral-3b-2410) for edge use; Mistral retired it on 31 December 2025 in favour of Ministral 3 3B.", "verified_on": "2026-10-12", "sources": [ "https://docs.mistral.ai/getting-started/models/models_overview/", "https://learn.microsoft.com/en-us/azure/foundry/foundry-models/concepts/models-from-partners" ], "unverified": [] }, { "id": "writer-palmyra-x", "name": "Palmyra X", "level": "family", "family_id": null, "developer": [ "writer" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Writer's enterprise language models for workflow automation, coding and data analysis, offered through Writer and Amazon Bedrock.", "verified_on": "2026-10-12", "sources": [ "https://dev.writer.com/home/models", "https://docs.aws.amazon.com/bedrock/latest/userguide/model-cards.html" ], "unverified": [] }, { "id": "writer-palmyra-x4", "name": "Palmyra X4", "level": "variant", "family_id": "writer-palmyra-x", "developer": [ "writer" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 128000, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Writer language model for complex tasks, workflow automation, coding and data analysis, with a 128k context; Writer lists it for deprecation on 14 December 2026.", "verified_on": "2026-10-12", "sources": [ "https://dev.writer.com/home/models", "https://docs.aws.amazon.com/bedrock/latest/userguide/model-cards.html" ], "unverified": [] }, { "id": "writer-palmyra-x5", "name": "Palmyra X5", "level": "variant", "family_id": "writer-palmyra-x", "developer": [ "writer" ], "model_class": "LLM", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": 1000000, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Writer language model with a 1M-token context, offered through Writer and Amazon Bedrock; Writer lists it for deprecation on 14 December 2026 in favour of Palmyra X6.", "verified_on": "2026-10-12", "sources": [ "https://dev.writer.com/home/models", "https://docs.aws.amazon.com/bedrock/latest/userguide/model-cards.html" ], "unverified": [] }, { "id": "openai-whisper-base", "name": "whisper-base", "level": "variant", "family_id": "openai-whisper", "developer": [ "openai" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "encoder-decoder transformer", "params_total": 72593920, "params_active": null, "params_checkpoint": null, "params_note": "Hugging Face safetensors count; an English-only base.en checkpoint is also published.", "context_length": null, "release_date": "2022-09-26", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Multilingual; the Hugging Face card lists Indonesian (id) among its language tags", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Base-size Whisper checkpoint for speech recognition and speech translation, trained on 680k hours of labelled audio.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openai/whisper", "https://huggingface.co/openai/whisper-base", "https://arxiv.org/abs/2212.04356" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "openai-whisper-medium", "name": "whisper-medium", "level": "variant", "family_id": "openai-whisper", "developer": [ "openai" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "encoder-decoder transformer", "params_total": 763857920, "params_active": null, "params_checkpoint": null, "params_note": "Hugging Face safetensors count; an English-only medium.en checkpoint is also published.", "context_length": null, "release_date": "2022-09-26", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Multilingual; the Hugging Face card lists Indonesian (id) among its language tags", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Medium-size Whisper checkpoint for speech recognition and speech translation, trained on 680k hours of labelled audio.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openai/whisper", "https://huggingface.co/openai/whisper-medium", "https://arxiv.org/abs/2212.04356" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "openai-whisper-large-v2", "name": "whisper-large-v2", "level": "variant", "family_id": "openai-whisper", "developer": [ "openai" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "encoder-decoder transformer", "params_total": 1543304960, "params_active": null, "params_checkpoint": null, "params_note": "Hugging Face safetensors count; same size as the original whisper-large (v1, September 2022).", "context_length": null, "release_date": "2022-12-05", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Multilingual; the Hugging Face card lists Indonesian (id) among its language tags", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Second version of the large Whisper checkpoint for multilingual speech recognition and speech translation; it replaced the original large model.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openai/whisper", "https://huggingface.co/openai/whisper-large-v2", "https://huggingface.co/openai/whisper-large", "https://arxiv.org/abs/2212.04356" ], "unverified": [ "summary", "indonesian_measured", "indonesia_hosting" ] }, { "id": "hugging-face-distil-whisper", "name": "Distil-Whisper", "level": "family", "family_id": null, "developer": [ "hugging-face" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "encoder-decoder transformer distilled from Whisper", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "166M (distil-small.en) to 756M (distil-large-v3 and v3.5); further checkpoints distil-medium.en and distil-large-v2.", "context_length": null, "release_date": "2023-11-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Knowledge-distilled versions of Whisper trained on pseudo-labelled open data, for faster speech recognition with close to Whisper accuracy.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2311.00430", "https://github.com/huggingface/distil-whisper", "https://huggingface.co/distil-whisper/distil-large-v3" ], "unverified": [] }, { "id": "hugging-face-distil-small-en", "name": "distil-small.en", "level": "variant", "family_id": "hugging-face-distil-whisper", "developer": [ "hugging-face" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 166132224, "params_active": null, "params_checkpoint": null, "params_note": "Smallest Distil-Whisper checkpoint (166M per the model card).", "context_length": null, "release_date": "2023-12-06", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server", "laptop" ], "footprint_note": "Model card recommends it for memory-constrained use.", "runtimes": [], "languages_note": "English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Smallest Distil-Whisper checkpoint for English speech recognition on memory-constrained hardware.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/distil-whisper/distil-small.en", "https://arxiv.org/abs/2311.00430" ], "unverified": [] }, { "id": "hugging-face-distil-large-v3", "name": "distil-large-v3", "level": "variant", "family_id": "hugging-face-distil-whisper", "developer": [ "hugging-face" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 756405760, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-03-21", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Third Distil-Whisper English release, a knowledge-distilled version of Whisper large-v3.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/distil-whisper/distil-large-v3", "https://arxiv.org/abs/2311.00430" ], "unverified": [] }, { "id": "hugging-face-distil-large-v3-5", "name": "distil-large-v3.5", "level": "variant", "family_id": "hugging-face-distil-whisper", "developer": [ "hugging-face" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 756405760, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-12-05", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Distil-Whisper checkpoint distilled from Whisper large-v3 on 98k hours of public data with a longer training schedule.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/distil-whisper/distil-large-v3.5", "https://arxiv.org/abs/2311.00430" ], "unverified": [] }, { "id": "kotoba-technologies-kotoba-whisper-v2-0", "name": "Kotoba-Whisper v2.0", "level": "model", "family_id": null, "developer": [ "kotoba-technologies" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 756405760, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-09-17", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Japanese", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Distilled Whisper model for Japanese speech recognition, made by Asahi Ushio and Kotoba Technologies.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/kotoba-tech/kotoba-whisper-v2.0" ], "unverified": [] }, { "id": "whisperx", "name": "WhisperX", "level": "model", "family_id": null, "developer": [ "university-of-oxford" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "Whisper with voice activity detection and phoneme forced alignment", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-03-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": "Code under the BSD 2-Clause licence (LICENSE file); no AURION licence record for it yet.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Long-form transcription system built on Whisper that adds voice activity detection and forced alignment for word-level timestamps.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2303.00747", "https://github.com/m-bain/whisperX" ], "unverified": [ "developer" ] }, { "id": "meta-hubert", "name": "HuBERT", "level": "family", "family_id": null, "developer": [ "meta" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": null, "architecture": "dense transformer", "architecture_detail": "self-supervised masked prediction of hidden units", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Base, Large and X-Large; Large also published fine-tuned on 960 h of LibriSpeech (hubert-large-ls960-ft).", "context_length": null, "release_date": "2021-06-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English (LibriSpeech and Libri-Light pre-training)", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Self-supervised speech representation model that learns by predicting masked hidden units; a backbone for speech recognition and other speech tasks.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2106.07447", "https://github.com/facebookresearch/fairseq/tree/main/examples/hubert", "https://huggingface.co/facebook/hubert-large-ls960-ft" ], "unverified": [] }, { "id": "meta-hubert-base", "name": "hubert-base-ls960", "level": "variant", "family_id": "meta-hubert", "developer": [ "meta" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": null, "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-06-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Base HuBERT model pre-trained on 16 kHz speech, to be fine-tuned for downstream speech tasks.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/facebook/hubert-base-ls960", "https://arxiv.org/abs/2106.07447" ], "unverified": [] }, { "id": "meta-hubert-large", "name": "hubert-large-ll60k", "level": "variant", "family_id": "meta-hubert", "developer": [ "meta" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": null, "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-06-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Large HuBERT model pre-trained on 16 kHz speech; a fine-tuned version for English speech recognition is published as hubert-large-ls960-ft.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/facebook/hubert-large-ll60k", "https://huggingface.co/facebook/hubert-large-ls960-ft", "https://arxiv.org/abs/2106.07447" ], "unverified": [] }, { "id": "meta-hubert-xlarge", "name": "hubert-xlarge-ll60k", "level": "variant", "family_id": "meta-hubert", "developer": [ "meta" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": null, "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-06-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Extra-large HuBERT model pre-trained on 16 kHz speech, meant to be fine-tuned for tasks such as speech recognition or speaker identification.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/facebook/hubert-xlarge-ll60k", "https://arxiv.org/abs/2106.07447" ], "unverified": [] }, { "id": "microsoft-wavlm", "name": "WavLM", "level": "family", "family_id": null, "developer": [ "microsoft" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": null, "architecture": "dense transformer", "architecture_detail": "self-supervised masked speech prediction with denoising", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Base, Base Plus and Large checkpoints.", "context_length": null, "release_date": "2021-10-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": "Licence taken from the microsoft/unilm repository LICENSE (MIT); the Hugging Face cards carry no licence tag.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Self-supervised speech model trained for full-stack speech processing, including speaker, paralinguistic and content tasks.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2110.13900", "https://github.com/microsoft/unilm/tree/master/wavlm", "https://huggingface.co/microsoft/wavlm-large" ], "unverified": [ "licence" ] }, { "id": "microsoft-wavlm-base-plus", "name": "wavlm-base-plus", "level": "variant", "family_id": "microsoft-wavlm", "developer": [ "microsoft" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": null, "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-10-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Base Plus WavLM model pre-trained on 16 kHz speech, to be fine-tuned for downstream speech tasks.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/microsoft/wavlm-base-plus", "https://arxiv.org/abs/2110.13900" ], "unverified": [ "licence" ] }, { "id": "microsoft-wavlm-large", "name": "wavlm-large", "level": "variant", "family_id": "microsoft-wavlm", "developer": [ "microsoft" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": null, "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-10-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Large WavLM model pre-trained on 16 kHz speech; used as an encoder for speaker verification, diarization and recognition.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/microsoft/wavlm-large", "https://arxiv.org/abs/2110.13900" ], "unverified": [ "licence" ] }, { "id": "nvidia-parakeet", "name": "Parakeet", "level": "family", "family_id": null, "developer": [ "nvidia" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "FastConformer encoder with CTC, RNN-T or TDT decoders", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "110M to 1.1B; checkpoints include parakeet-ctc-0.6b, parakeet-ctc-1.1b, parakeet-rnnt-0.6b, parakeet-rnnt-1.1b, parakeet-tdt-1.1b, parakeet-tdt_ctc-110m, parakeet-tdt_ctc-0.6b-ja, parakeet-unified-en-0.6b. The Parakeet-TDT-0.6B line has its own AURION family.", "context_length": null, "release_date": "2023-12-28", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Mostly English; one Japanese checkpoint", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA NeMo family of FastConformer speech recognition models with CTC, transducer and TDT decoders.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/nvidia/parakeet-ctc-1.1b", "https://docs.nvidia.com/nemo-framework/user-guide/latest/nemotoolkit/asr/all_chkpt.html" ], "unverified": [] }, { "id": "nvidia-parakeet-ctc-1-1b", "name": "parakeet-ctc-1.1b", "level": "variant", "family_id": "nvidia-parakeet", "developer": [ "nvidia" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": 1062626305, "params_active": null, "params_checkpoint": null, "params_note": "XXL FastConformer CTC, about 1.1B per the card; a 0.6B CTC checkpoint also exists.", "context_length": null, "release_date": "2023-12-28", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Large FastConformer CTC model that transcribes English speech into lowercase text.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/nvidia/parakeet-ctc-1.1b", "https://huggingface.co/nvidia/parakeet-ctc-0.6b" ], "unverified": [] }, { "id": "nvidia-parakeet-rnnt-1-1b", "name": "parakeet-rnnt-1.1b", "level": "variant", "family_id": "nvidia-parakeet", "developer": [ "nvidia" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": 1070518785, "params_active": null, "params_checkpoint": null, "params_note": "XXL FastConformer Transducer, about 1.1B per the card.", "context_length": null, "release_date": "2023-12-27", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Large FastConformer transducer (RNN-T) model for English speech recognition.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/nvidia/parakeet-rnnt-1.1b" ], "unverified": [] }, { "id": "nvidia-parakeet-tdt-1-1b", "name": "parakeet-tdt-1.1b", "level": "variant", "family_id": "nvidia-parakeet", "developer": [ "nvidia" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "XXL FastConformer TDT, about 1.1B per the card.", "context_length": null, "release_date": "2024-01-25", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Large FastConformer model with a token-and-duration transducer (TDT) decoder for English speech recognition.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/nvidia/parakeet-tdt-1.1b" ], "unverified": [] }, { "id": "nvidia-parakeet-tdt-ctc-110m", "name": "parakeet-tdt_ctc-110m", "level": "variant", "family_id": "nvidia-parakeet", "developer": [ "nvidia" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": 114000000, "params_active": null, "params_checkpoint": null, "params_note": "About 114M per the card.", "context_length": null, "release_date": "2024-09-17", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Smallest Parakeet checkpoint, a hybrid FastConformer TDT-CTC model for English speech recognition.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/nvidia/parakeet-tdt_ctc-110m" ], "unverified": [] }, { "id": "nvidia-parakeet-unified-en-0-6b", "name": "parakeet-unified-en-0.6b", "level": "variant", "family_id": "nvidia-parakeet", "developer": [ "nvidia" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": 600000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04-07", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "nvidia-open-model-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "English transducer model that combines offline and streaming recognition (minimum latency 160 ms) in one model.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/nvidia/parakeet-unified-en-0.6b" ], "unverified": [] }, { "id": "nvidia-multitalker-parakeet-streaming-0-6b-v1", "name": "multitalker-parakeet-streaming-0.6b-v1", "level": "variant", "family_id": "nvidia-parakeet", "developer": [ "nvidia" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": 600000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10-15", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "nvidia-open-model-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Streaming FastConformer model for transcribing overlapping speech from several speakers.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/nvidia/multitalker-parakeet-streaming-0.6b-v1" ], "unverified": [] }, { "id": "nvidia-canary-1b-v2", "name": "canary-1b-v2", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "FastConformer encoder with Transformer decoder", "params_total": 978998272, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-08-04", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "25 languages in the card language tags (bg, cs, da and others)", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multilingual Canary model from NVIDIA NeMo for speech recognition in 25 languages.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/nvidia/canary-1b-v2" ], "unverified": [] }, { "id": "nvidia-canary-180m-flash", "name": "canary-180m-flash", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "FastConformer encoder with Transformer decoder", "params_total": 182000000, "params_active": null, "params_checkpoint": null, "params_note": "182 million parameters per the card.", "context_length": null, "release_date": "2025-03-11", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English, German, French, Spanish", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small multilingual multitask Canary Flash model for speech recognition and translation, with experimental word timestamps.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/nvidia/canary-180m-flash" ], "unverified": [] }, { "id": "nvidia-nemotron-speech-streaming-en-0-6b", "name": "Nemotron Speech Streaming EN 0.6B", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "cache-aware FastConformer RNN-T", "params_total": 600000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "nvidia-open-model-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "English streaming speech recognition model for both low-latency streaming and high-throughput batch transcription.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/nvidia/nemotron-speech-streaming-en-0.6b", "https://build.nvidia.com/nvidia/nemotron-asr-streaming/" ], "unverified": [] }, { "id": "nvidia-nemotron-3-5-asr-streaming-0-6b", "name": "Nemotron 3.5 ASR Streaming 0.6B", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "cache-aware FastConformer RNN-T with language-ID prompting", "params_total": 637997088, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-06-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "openmdw-1-1", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "40 language-locales (32 ready out of the box, 8 need fine-tuning); Indonesian is not in the listed locales", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multilingual extension of Nemotron Speech Streaming that transcribes 40 language-locales with optional automatic language detection.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/nvidia/nemotron-3.5-asr-streaming-0.6b" ], "unverified": [] }, { "id": "nvidia-fastconformer", "name": "FastConformer", "level": "family", "family_id": null, "developer": [ "nvidia" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "Conformer redesigned with 8x downsampling and linearly scalable attention", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "NeMo publishes CTC, transducer and hybrid checkpoints in about 25 languages (stt__fastconformer_*); Large is about 114M.", "context_length": null, "release_date": "2023-05-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Redesigned Conformer encoder that is 2.8 times faster and is the basis of the NeMo Parakeet and Canary models.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2305.05084", "https://huggingface.co/nvidia/stt_en_fastconformer_transducer_large" ], "unverified": [] }, { "id": "nvidia-stt-en-fastconformer-transducer-large", "name": "stt_en_fastconformer_transducer_large", "level": "variant", "family_id": "nvidia-fastconformer", "developer": [ "nvidia" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": 114000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-06-08", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Large FastConformer transducer model for English speech recognition.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/nvidia/stt_en_fastconformer_transducer_large" ], "unverified": [] }, { "id": "google-conformer", "name": "Conformer", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "convolution-augmented transformer encoder", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "NVIDIA NeMo publishes Conformer-CTC and Conformer-Transducer checkpoints (stt_en_conformer_ctc_large is about 120M).", "context_length": null, "release_date": "2020-05-16", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Speech recognition encoder that combines self-attention and convolution; widely reimplemented, for example as NeMo Conformer-CTC checkpoints.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2005.08100", "https://huggingface.co/nvidia/stt_en_conformer_ctc_large" ], "unverified": [ "developer" ] }, { "id": "google-contextnet", "name": "ContextNet", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "CNN", "architecture_detail": "CNN-RNN-transducer with squeeze-and-excitation", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-05-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fully convolutional encoder with global context for transducer speech recognition; NeMo publishes ContextNet checkpoints.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2005.03191" ], "unverified": [ "developer" ] }, { "id": "squeezeformer", "name": "Squeezeformer", "level": "model", "family_id": null, "developer": [ "uc-berkeley" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "efficient Conformer variant", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-06-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Efficient transformer encoder for speech recognition that revises Conformer design choices; NeMo publishes Squeezeformer checkpoints.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2206.00888" ], "unverified": [ "developer" ] }, { "id": "nvidia-citrinet", "name": "Citrinet", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "CNN", "architecture_detail": "1D time-channel separable convolutions with squeeze-and-excitation, CTC", "params_total": 145000000, "params_active": null, "params_checkpoint": null, "params_note": "stt_en_citrinet_1024_gamma_0_25 is about 140M to 145M.", "context_length": null, "release_date": "2021-04-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Convolutional CTC speech recognition model from NVIDIA NeMo.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2104.01721", "https://huggingface.co/nvidia/stt_en_citrinet_1024_gamma_0_25" ], "unverified": [] }, { "id": "nvidia-quartznet", "name": "QuartzNet", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "CNN", "architecture_detail": "1D time-channel separable convolutions, CTC", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "QuartzNet 15x5 is the common English checkpoint.", "context_length": null, "release_date": "2019-10-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Compact convolutional CTC speech recognition model; QuartzNet 15x5 is also in the OpenVINO model zoo.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1910.10261", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/quartznet-15x5-en" ], "unverified": [] }, { "id": "nvidia-jasper", "name": "Jasper", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "CNN", "architecture_detail": "1D convolutions with residual connections, CTC", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-04-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "End-to-end convolutional acoustic model for speech recognition, trained with the NovoGrad optimiser.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1904.03288", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/dle/models/jasper__pyt_ckpt" ], "unverified": [] }, { "id": "nvidia-marblenet", "name": "MarbleNet", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "audio classification", "modalities_in": [ "audio" ], "modalities_out": null, "architecture": "CNN", "architecture_detail": "1D time-channel separable convolutions", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Frame-VAD Multilingual MarbleNet v2.0 is the current checkpoint.", "context_length": null, "release_date": "2020-10-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "nvidia-open-model-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small convolutional voice activity detection model used as the first step for speech recognition and diarization.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2010.13886", "https://huggingface.co/nvidia/Frame_VAD_Multilingual_MarbleNet_v2.0" ], "unverified": [] }, { "id": "nvidia-matchboxnet", "name": "MatchboxNet", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "audio classification", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": "1D time-channel separable convolutions", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-04-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small convolutional model for speech command recognition, evaluated on Google Speech Commands.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2004.08531", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/nemo/models/commandrecognition_en_matchboxnet3x1x64_v1" ], "unverified": [] }, { "id": "nvidia-titanet", "name": "TitaNet", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "speaker", "modalities_in": [ "audio" ], "modalities_out": null, "architecture": "CNN", "architecture_detail": "1D depth-wise separable convolutions with squeeze-and-excitation", "params_total": 23000000, "params_active": null, "params_checkpoint": null, "params_note": "TitaNet-Large is about 23M.", "context_length": null, "release_date": "2021-10-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Speaker embedding model for speaker verification and diarization.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2110.04410", "https://huggingface.co/nvidia/speakerverification_en_titanet_large" ], "unverified": [] }, { "id": "nvidia-msdd", "name": "MSDD (multi-scale diarization decoder)", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "speaker", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": "multi-scale speaker diarization with dynamic scale weighting", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-03-30", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multi-scale speaker diarization model from NVIDIA NeMo; a telephonic checkpoint is published on NGC.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2203.15974", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/nemo/models/diar_msdd_telephonic" ], "unverified": [] }, { "id": "mistral-ai-voxtral-mini-3b-2507", "name": "Voxtral Mini 3B 2507", "level": "model", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "speech recognition", "modalities_in": [ "audio", "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 4676271104, "params_active": null, "params_checkpoint": null, "params_note": "Built on Ministral 3B; 4.68B in the Hugging Face safetensors count.", "context_length": null, "release_date": "2025-07-01", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Ministral 3B with audio input, for speech transcription, translation and audio understanding.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/mistralai/Voxtral-Mini-3B-2507", "https://mistral.ai/news/voxtral", "https://arxiv.org/abs/2507.13264" ], "unverified": [] }, { "id": "mistral-ai-voxtral-mini-4b-realtime-2602", "name": "Voxtral Mini 4B Realtime 2602", "level": "model", "family_id": null, "developer": [ "mistral-ai" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 4429679360, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01-21", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "laptop" ], "footprint_note": "Card says it is optimised for on-device deployment.", "runtimes": [], "languages_note": "13 languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multilingual real-time transcription model with under 500 ms delay, for voice assistants and live subtitles.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/mistralai/Voxtral-Mini-4B-Realtime-2602" ], "unverified": [] }, { "id": "microsoft-vibevoice-asr", "name": "VibeVoice-ASR", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": 8674021857, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-01-21", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Multilingual; the card lists Indonesian (id)", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft VibeVoice speech recognition model released with open weights.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/microsoft/VibeVoice-ASR", "https://github.com/microsoft/VibeVoice" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "microsoft-mai-transcribe-1", "name": "MAI-Transcribe-1", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "gated preview" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "25 languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "preview", "summary": "Microsoft AI multilingual speech-to-text model, in public preview on Microsoft Foundry at launch.", "verified_on": "2026-10-12", "sources": [ "https://microsoft.ai/news/state-of-the-art-speech-recognition-with-mai-transcribe-1/" ], "unverified": [] }, { "id": "microsoft-mai-transcribe-2", "name": "MAI-Transcribe-2", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-09-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "60 languages", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft AI transcription model, available through Microsoft Foundry, the MAI Playground and OpenRouter.", "verified_on": "2026-10-12", "sources": [ "https://microsoft.ai/news/mai-transcribe-2-is-the-fastest-most-accurate-and-cheapest-speech-recognition-model-in-the-world/", "https://microsoft.ai/models/mai-transcribe-2/" ], "unverified": [] }, { "id": "openai-gpt-4o-transcribe", "name": "gpt-4o-transcribe", "level": "model", "family_id": null, "developer": [ "openai" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "A smaller GPT-4o Mini Transcribe is listed alongside it.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenAI speech-to-text model powered by GPT-4o, served through the transcription endpoint.", "verified_on": "2026-10-12", "sources": [ "https://developers.openai.com/api/docs/models/gpt-4o-transcribe" ], "unverified": [] }, { "id": "carnegie-mellon-university-owsm-v4", "name": "OWSM v4", "level": "family", "family_id": null, "developer": [ "carnegie-mellon-university" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Includes owsm_v4_medium_1B (1B) and the CTC model owsm_ctc_v4_1B.", "context_length": null, "release_date": "2025-01-16", "release_date_basis": "registry date", "weights_date": null, "openness": "fully open", "licence": "cc-by-4-0", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open Whisper-style Speech Models from the ESPnet team, trained and released with open data and code.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/espnet/owsm_ctc_v4_1B", "https://huggingface.co/espnet/owsm_v4_medium_1B", "https://arxiv.org/abs/2309.13876" ], "unverified": [] }, { "id": "carnegie-mellon-university-owsm-ctc-v4-1b", "name": "OWSM-CTC v4 1B", "level": "variant", "family_id": "carnegie-mellon-university-owsm-v4", "developer": [ "carnegie-mellon-university" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": 1000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-01-16", "release_date_basis": "registry date", "weights_date": null, "openness": "fully open", "licence": "cc-by-4-0", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Non-autoregressive CTC version of OWSM v4 for multilingual speech recognition.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/espnet/owsm_ctc_v4_1B" ], "unverified": [ "params_total" ] }, { "id": "dataocean-ai-dolphin", "name": "Dolphin", "level": "model", "family_id": null, "developer": [ "dataocean-ai" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "Whisper-style architecture", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "dolphin-small and dolphin-base are public.", "context_length": null, "release_date": "2025-03-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Eastern languages (paper title)", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Large multilingual speech recognition model that extends the Whisper architecture to more Eastern languages.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2503.20212", "https://github.com/DataoceanAI/Dolphin", "https://huggingface.co/DataoceanAI/dolphin-small" ], "unverified": [] }, { "id": "k2-fsa-zipformer", "name": "Zipformer", "level": "model", "family_id": null, "developer": [ "k2-fsa" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "U-Net-like transformer encoder", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-10-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Faster and more memory-efficient transformer encoder for speech recognition, from Next-gen Kaldi (icefall).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2310.11230", "https://github.com/k2-fsa/icefall" ], "unverified": [] }, { "id": "baidu-deep-speech-2", "name": "Deep Speech 2", "level": "model", "family_id": null, "developer": [ "baidu" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": "end-to-end RNN with CTC", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2015-12-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "End-to-end deep learning speech recognition for English and Mandarin, replacing hand-engineered pipelines.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1512.02595", "https://github.com/SeanNaren/deepspeech.pytorch" ], "unverified": [ "developer" ] }, { "id": "mozilla-deepspeech", "name": "Mozilla DeepSpeech", "level": "model", "family_id": null, "developer": [ "mozilla" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open source", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": "Code under MPL 2.0 (LICENSE file); versions 0.6.1 and 0.8.2 are in the OpenVINO model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open-source speech-to-text engine based on Baidu's Deep Speech paper.", "verified_on": "2026-10-12", "sources": [ "https://github.com/mozilla/DeepSpeech", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/mozilla-deepspeech-0.8.2" ], "unverified": [] }, { "id": "alibaba-funasr-conformer-en", "name": "FunASR Conformer (English)", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "speech recognition", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-02-01", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "funasr-model-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "English Conformer speech recognition model in the FunASR toolkit.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/funasr/conformer-en", "https://github.com/modelscope/FunASR" ], "unverified": [] }, { "id": "alibaba-fsmn-vad", "name": "FSMN-VAD", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "audio classification", "modalities_in": [ "audio" ], "modalities_out": null, "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-02-01", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Voice activity detection model from the FunASR toolkit.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/funasr/fsmn-vad", "https://www.modelscope.cn/models/iic/speech_fsmn_vad_zh-cn-16k-common-pytorch" ], "unverified": [] }, { "id": "rednote-firered-vad", "name": "FireRedVAD", "level": "model", "family_id": null, "developer": [ "rednote" ], "model_class": "audio classification", "modalities_in": [ "audio" ], "modalities_out": null, "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-02-11", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Voice activity detection model from the FireRed team.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/FireRedTeam/FireRedVAD" ], "unverified": [ "developer" ] }, { "id": "pipecat-smart-turn-v3", "name": "Smart Turn v3", "level": "model", "family_id": null, "developer": [ "pipecat" ], "model_class": "audio classification", "modalities_in": [ "audio" ], "modalities_out": null, "architecture": "other", "architecture_detail": null, "params_total": 8000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-09-11", "release_date_basis": "registry date", "weights_date": null, "openness": "open source", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "BSD 2-Clause per the Hugging Face tag; no AURION licence record for it yet.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Semantic voice activity model that decides from the raw waveform whether a speaker has finished their turn.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/pipecat-ai/smart-turn-v3", "https://www.daily.co/blog/announcing-smart-turn-v3-with-cpu-inference-in-just-12ms/" ], "unverified": [] }, { "id": "pyannoteai-speaker-diarization-3-1", "name": "pyannote speaker-diarization-3.1", "level": "model", "family_id": null, "developer": [ "pyannoteai" ], "model_class": "speaker", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-11-16", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download", "gated preview" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open speaker diarization pipeline from pyannote.audio, built on the segmentation-3.0 model.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/pyannote/speaker-diarization-3.1" ], "unverified": [] }, { "id": "pyannoteai-segmentation-3-0", "name": "pyannote segmentation-3.0", "level": "model", "family_id": null, "developer": [ "pyannoteai" ], "model_class": "speaker", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-09-22", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download", "gated preview" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Speaker segmentation model used inside pyannote diarization pipelines.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/pyannote/segmentation-3.0" ], "unverified": [] }, { "id": "brno-university-of-technology-diarizen", "name": "DiariZen (WavLM Large s80 md v2)", "level": "model", "family_id": null, "developer": [ "brno-university-of-technology" ], "model_class": "speaker", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": "EEND with pruned WavLM Large and Conformer layers", "params_total": 63300000, "params_active": null, "params_checkpoint": null, "params_note": "WavLM Large pruned from 316.6M to 63.3M parameters.", "context_length": null, "release_date": "2025-12-09", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Speaker diarization model trained on far-field single-channel audio, using a structurally pruned WavLM Large.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/BUT-FIT/diarizen-wavlm-large-s80-md-v2", "https://github.com/BUTSpeechFIT/DiariZen" ], "unverified": [] }, { "id": "alibaba-eres2net", "name": "ERes2Net", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "speaker", "modalities_in": [ "audio" ], "modalities_out": null, "architecture": "CNN", "architecture_detail": "Enhanced Res2Net with local and global feature fusion", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "ERes2Net and ERes2NetV2 checkpoints are on ModelScope in the 3D-Speaker project.", "context_length": null, "release_date": "2023-05-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Speaker verification model that fuses multi-scale features for speaker embeddings.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2305.12838", "https://github.com/modelscope/3D-Speaker" ], "unverified": [] }, { "id": "alibaba-emotion2vec-plus-large", "name": "emotion2vec+ large", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "audio classification", "modalities_in": [ "audio" ], "modalities_out": [], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-05-15", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "funasr-model-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Speech emotion recognition foundation model trained to work across languages and recording conditions.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/emotion2vec/emotion2vec_plus_large", "https://arxiv.org/abs/2312.15185", "https://www.modelscope.cn/models/iic/emotion2vec_plus_large" ], "unverified": [ "developer" ] }, { "id": "audeering-wav2vec2-emotion-msp-dim", "name": "wav2vec2-large-robust-12-ft-emotion-msp-dim", "level": "model", "family_id": null, "developer": [ "audeering" ], "model_class": "audio classification", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": 165336707, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-03-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-sa-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Pruned wav2vec 2.0 model fine-tuned on MSP-Podcast that predicts arousal, dominance and valence from speech.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/audeering/wav2vec2-large-robust-12-ft-emotion-msp-dim", "https://arxiv.org/abs/2203.07378" ], "unverified": [] }, { "id": "google-trillsson", "name": "TRILLsson", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "embedding", "modalities_in": [ "audio" ], "modalities_out": null, "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "FRILL is a related mobile-sized paralinguistic embedding model (MobileNet v3).", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Paralinguistic speech embedding models distilled from a larger model, for tasks such as emotion and language identification.", "verified_on": "2026-10-12", "sources": [ "https://github.com/tensorflow/tfhub.dev/blob/master/assets/docs/google/models/trillsson1/1.md", "https://www.kaggle.com/models/google/trillsson", "https://github.com/tensorflow/tfhub.dev/blob/master/assets/docs/google/models/nonsemantic-speech-benchmark/frill/1.md" ], "unverified": [ "licence" ] }, { "id": "speechbrain-sepformer", "name": "SepFormer", "level": "model", "family_id": null, "developer": [ "speechbrain" ], "model_class": "music and audio", "modalities_in": [ "audio" ], "modalities_out": [ "audio" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "About 11 SpeechBrain checkpoints for separation and enhancement (WSJ0-2Mix, Libri2Mix, WHAMR and others).", "context_length": null, "release_date": "2020-10-25", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Transformer model for speech separation that splits overlapping speakers into separate audio tracks.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2010.13154", "https://huggingface.co/speechbrain/sepformer-wsj02mix" ], "unverified": [] }, { "id": "speechbrain-metricgan-plus", "name": "MetricGAN+", "level": "model", "family_id": null, "developer": [ "speechbrain" ], "model_class": "music and audio", "modalities_in": [ "audio" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": "GAN trained against a learned perceptual metric", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-04-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Speech enhancement model trained to optimise a perceptual quality metric; SpeechBrain publishes a VoiceBank checkpoint.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2104.03538", "https://huggingface.co/speechbrain/metricgan-plus-voicebank" ], "unverified": [] }, { "id": "speechbrain-spkrec-xvect-voxceleb", "name": "x-vector speaker recognition (VoxCeleb)", "level": "model", "family_id": null, "developer": [ "speechbrain" ], "model_class": "speaker", "modalities_in": [ "audio" ], "modalities_out": null, "architecture": "other", "architecture_detail": "TDNN x-vector", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "TDNN speaker embedding model trained on VoxCeleb 1 and 2 for speaker recognition.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/speechbrain/spkrec-xvect-voxceleb" ], "unverified": [] }, { "id": "speechbrain-lang-id-voxlingua107-ecapa", "name": "VoxLingua107 language ID (ECAPA)", "level": "model", "family_id": null, "developer": [ "speechbrain" ], "model_class": "audio classification", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": "ECAPA-TDNN", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "107 languages including Indonesian", "indonesian_support": "listed by developer", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Spoken language identification model trained on VoxLingua107 with the ECAPA-TDNN architecture.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/speechbrain/lang-id-voxlingua107-ecapa" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "speechbrain-emotion-recognition-wav2vec2-iemocap", "name": "Emotion recognition wav2vec2 (IEMOCAP)", "level": "model", "family_id": null, "developer": [ "speechbrain" ], "model_class": "audio classification", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Speech emotion recognition with a fine-tuned wav2vec2 base model on IEMOCAP.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/speechbrain/emotion-recognition-wav2vec2-IEMOCAP" ], "unverified": [] }, { "id": "speechbrain-vad-crdnn-libriparty", "name": "VAD CRDNN (LibriParty)", "level": "model", "family_id": null, "developer": [ "speechbrain" ], "model_class": "audio classification", "modalities_in": [ "audio" ], "modalities_out": null, "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open source", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Voice activity detection model trained on LibriParty that outputs the segments containing speech.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/speechbrain/vad-crdnn-libriparty" ], "unverified": [] }, { "id": "kakao-enterprise-vits", "name": "VITS", "level": "model", "family_id": null, "developer": [ "kakao-enterprise" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": "conditional VAE with adversarial learning", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-06-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "End-to-end text-to-speech model that generates audio in one stage; the basis of many later TTS models.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2106.06103", "https://github.com/jaywalnut310/vits" ], "unverified": [ "developer" ] }, { "id": "kakao-enterprise-glow-tts", "name": "Glow-TTS", "level": "model", "family_id": null, "developer": [ "kakao-enterprise" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": "flow-based with monotonic alignment search", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-05-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Flow-based parallel text-to-speech model that learns its own alignment without an external aligner.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2005.11129" ], "unverified": [ "developer" ] }, { "id": "kakao-enterprise-hifi-gan", "name": "HiFi-GAN", "level": "model", "family_id": null, "developer": [ "kakao-enterprise" ], "model_class": "speech synthesis", "modalities_in": [ "audio" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": "GAN vocoder", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-10-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "GAN vocoder that turns mel-spectrograms into high-fidelity audio efficiently; widely used in TTS pipelines (NeMo, SpeechBrain).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2010.05646", "https://github.com/jik876/hifi-gan", "https://huggingface.co/speechbrain/tts-hifigan-libritts-22050Hz", "https://huggingface.co/nvidia/tts_hifigan" ], "unverified": [ "developer" ] }, { "id": "coqui-yourtts", "name": "YourTTS", "level": "model", "family_id": null, "developer": [ "coqui" ], "model_class": "speech synthesis", "modalities_in": [ "text", "audio" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": "builds on VITS", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-12-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": "Coqui TTS code under MPL 2.0 (LICENSE file).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Zero-shot multi-speaker and multilingual TTS and voice conversion model built on VITS.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2112.02418", "https://github.com/coqui-ai/TTS" ], "unverified": [] }, { "id": "james-betker-tortoise-tts", "name": "Tortoise TTS", "level": "model", "family_id": null, "developer": [ "james-betker" ], "model_class": "speech synthesis", "modalities_in": [ "text", "audio" ], "modalities_out": [ "audio" ], "architecture": "autoregressive", "architecture_detail": "autoregressive transformer with diffusion decoder", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-05-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multi-voice text-to-speech system that applies image-generation methods (autoregressive transformers and diffusion) to speech.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2305.07243", "https://github.com/neonbjb/tortoise-tts" ], "unverified": [] }, { "id": "google-tacotron-2", "name": "Tacotron 2", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": "sequence-to-sequence mel predictor with WaveNet vocoder", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Public checkpoints come from NVIDIA (BSD-3-Clause) and SpeechBrain (LJSpeech).", "context_length": null, "release_date": "2017-12-16", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Text-to-speech architecture that predicts mel-spectrograms from characters and voices them with a WaveNet vocoder.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1712.05884", "https://github.com/NVIDIA/tacotron2", "https://huggingface.co/speechbrain/tts-tacotron2-ljspeech" ], "unverified": [ "developer" ] }, { "id": "microsoft-fastspeech-2", "name": "FastSpeech 2", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "dense transformer", "architecture_detail": "non-autoregressive", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "ESPnet publishes a FastSpeech2Conformer variant.", "context_length": null, "release_date": "2020-06-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fast non-autoregressive text-to-speech model trained without a teacher model.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2006.04558", "https://huggingface.co/espnet/fastspeech2_conformer" ], "unverified": [ "developer" ] }, { "id": "microsoft-speecht5", "name": "SpeechT5", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "speech synthesis", "modalities_in": [ "text", "audio" ], "modalities_out": [ "audio", "text" ], "architecture": "dense transformer", "architecture_detail": "shared encoder-decoder with modal pre and post nets", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-10-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Unified speech and text encoder-decoder model; published checkpoints cover text-to-speech, voice conversion and recognition.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2110.07205", "https://huggingface.co/microsoft/speecht5_tts" ], "unverified": [] }, { "id": "microsoft-vall-e", "name": "VALL-E", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "speech synthesis", "modalities_in": [ "text", "audio" ], "modalities_out": [ "audio" ], "architecture": "autoregressive", "architecture_detail": "neural codec language model", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-01-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "research only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Text-to-speech approach that treats synthesis as language modelling over neural codec tokens, enabling zero-shot voice prompts.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2301.02111" ], "unverified": [ "developer" ] }, { "id": "nvidia-fastpitch", "name": "FastPitch", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 45000000, "params_active": null, "params_checkpoint": null, "params_note": "NeMo tts_en_fastpitch is about 45M.", "context_length": null, "release_date": "2020-06-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Parallel text-to-speech model conditioned on predicted pitch contours, used in NeMo and Riva.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2006.06873", "https://huggingface.co/nvidia/tts_en_fastpitch" ], "unverified": [] }, { "id": "nvidia-waveglow", "name": "WaveGlow", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "speech synthesis", "modalities_in": [ "audio" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": "flow-based vocoder", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-10-31", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "bsd-3-clause", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Flow-based vocoder that generates speech from mel-spectrograms without autoregression.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1811.00002", "https://github.com/NVIDIA/waveglow" ], "unverified": [] }, { "id": "nvidia-flowtron", "name": "Flowtron", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "autoregressive", "architecture_detail": "autoregressive flow", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-05-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Autoregressive flow-based text-to-speech model with control over speech variation and style transfer.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2005.05957", "https://github.com/NVIDIA/flowtron" ], "unverified": [] }, { "id": "nvidia-bigvgan", "name": "BigVGAN", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "speech synthesis", "modalities_in": [ "audio" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": "universal GAN vocoder", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "BigVGAN v2 checkpoints (July 2024) cover 22, 24 and 44 kHz.", "context_length": null, "release_date": "2022-06-09", "release_date_basis": "announcement page", "weights_date": "2024-07-15", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Universal neural vocoder for high-fidelity audio across many speakers and recording conditions.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2206.04658", "https://github.com/NVIDIA/BigVGAN", "https://huggingface.co/nvidia/bigvgan_v2_44khz_128band_512x" ], "unverified": [] }, { "id": "nvidia-magpie-tts-multilingual-357m", "name": "Magpie TTS Multilingual 357M", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "dense transformer", "architecture_detail": "encoder-decoder transformer", "params_total": 364000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-12-11", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "nvidia-open-model-license", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "ar, de, en, es, fr, hi, it, ja, ko, pt, vi, zh", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multilingual text-to-speech model with five English speaker voices, also offered in Riva.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/nvidia/magpie_tts_multilingual_357m" ], "unverified": [] }, { "id": "google-deepmind-wavenet", "name": "WaveNet", "level": "model", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "speech synthesis", "modalities_in": [ "audio" ], "modalities_out": [ "audio" ], "architecture": "autoregressive", "architecture_detail": "dilated causal convolutions", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2016-09-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Autoregressive generative model of raw audio waveforms, used as a vocoder for speech synthesis.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1609.03499" ], "unverified": [ "developer" ] }, { "id": "google-deepmind-wavernn", "name": "WaveRNN", "level": "model", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "speech synthesis", "modalities_in": [ "audio" ], "modalities_out": [ "audio" ], "architecture": "autoregressive", "architecture_detail": "single-layer RNN", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-02-23", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Efficient recurrent neural audio synthesis model for fast text-to-speech sampling; in the OpenVINO model zoo.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1802.08435", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/wavernn" ], "unverified": [ "developer" ] }, { "id": "descript-melgan", "name": "MelGAN", "level": "model", "family_id": null, "developer": [ "descript" ], "model_class": "speech synthesis", "modalities_in": [ "audio" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": "GAN vocoder", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-10-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "GAN-based waveform generator for conditional audio synthesis from mel-spectrograms.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1910.06711", "https://github.com/descriptinc/melgan-neurips" ], "unverified": [] }, { "id": "naver-parallel-wavegan", "name": "Parallel WaveGAN", "level": "model", "family_id": null, "developer": [ "naver" ], "model_class": "speech synthesis", "modalities_in": [ "audio" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": "non-autoregressive WaveNet trained with GAN and spectrogram losses", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-10-25", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fast, small distillation-free waveform generator trained with adversarial and multi-resolution spectrogram losses.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1910.11480", "https://github.com/kan-bayashi/ParallelWaveGAN" ], "unverified": [ "developer" ] }, { "id": "baidu-diffwave", "name": "DiffWave", "level": "model", "family_id": null, "developer": [ "baidu" ], "model_class": "speech synthesis", "modalities_in": [ "audio" ], "modalities_out": [ "audio" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-09-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Diffusion model for conditional and unconditional waveform generation, used as a vocoder.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2009.09761", "https://github.com/lmnt-com/diffwave" ], "unverified": [ "developer" ] }, { "id": "columbia-university-styletts-2", "name": "StyleTTS 2", "level": "model", "family_id": null, "developer": [ "columbia-university" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "diffusion", "architecture_detail": "style diffusion with adversarial training using speech language models", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-06-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Text-to-speech model that samples speaking style through diffusion to approach human-level naturalness.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2306.07691", "https://github.com/yl4579/StyleTTS2" ], "unverified": [ "developer" ] }, { "id": "myshell-openvoice", "name": "OpenVoice", "level": "model", "family_id": null, "developer": [ "myshell", "mit" ], "model_class": "speech synthesis", "modalities_in": [ "text", "audio" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "OpenVoice V2 (April 2024) improves audio quality.", "context_length": null, "release_date": "2023-12-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Voice cloning model that copies a voice from a short clip and speaks in several languages with style control.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2312.01479", "https://github.com/myshell-ai/OpenVoice", "https://huggingface.co/myshell-ai/OpenVoiceV2" ], "unverified": [] }, { "id": "myshell-melotts", "name": "MeloTTS", "level": "model", "family_id": null, "developer": [ "myshell", "mit" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-02-29", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": "English accents, Spanish, French, Chinese, Japanese, Korean", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multilingual text-to-speech library by MIT and MyShell.ai; also packaged for Qualcomm AI Hub.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/myshell-ai/MeloTTS-English", "https://github.com/myshell-ai/MeloTTS", "https://aihub.qualcomm.com/models" ], "unverified": [ "languages_note" ] }, { "id": "rhasspy-piper", "name": "Piper", "level": "model", "family_id": null, "developer": [ "rhasspy" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": "VITS-based voices exported to ONNX", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-06-22", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server", "mobile or NPU", "offline capable" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "Local text-to-speech system with many downloadable voices; the original repository is discontinued and development moved to OHF-Voice/piper1-gpl.", "verified_on": "2026-10-12", "sources": [ "https://github.com/rhasspy/piper", "https://huggingface.co/rhasspy/piper-voices", "https://github.com/OHF-Voice/piper1-gpl" ], "unverified": [ "architecture_detail" ] }, { "id": "silero-tts", "name": "Silero TTS", "level": "model", "family_id": null, "developer": [ "silero" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Released in versions up to V5, with Turkic, Caucasian and CIS speaker sets.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-sa-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "CPU server" ], "footprint_note": "The repository LICENSE file is CC BY-NC-SA 4.0; the README says some models are MIT.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Compact multilingual text-to-speech models from Silero, loadable through PyTorch Hub.", "verified_on": "2026-10-12", "sources": [ "https://github.com/snakers4/silero-models", "https://pytorch.org/hub/snakers4_silero-models_tts" ], "unverified": [ "licence" ] }, { "id": "canopy-labs-orpheus-tts", "name": "Orpheus TTS", "level": "model", "family_id": null, "developer": [ "canopy-labs" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "dense transformer", "architecture_detail": "Llama 3B backbone", "params_total": 3782986752, "params_active": null, "params_checkpoint": null, "params_note": "Multilingual research models followed in April 2025.", "context_length": null, "release_date": "2025-03-17", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open text-to-speech system built on a Llama 3B language model backbone.", "verified_on": "2026-10-12", "sources": [ "https://github.com/canopyai/Orpheus-TTS", "https://huggingface.co/canopylabs/orpheus-3b-0.1-ft" ], "unverified": [ "licence" ] }, { "id": "2noise-chattts", "name": "ChatTTS", "level": "model", "family_id": null, "developer": [ "2noise" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-05-25", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": "Code under AGPL 3.0 (LICENSE file); weights CC BY-NC 4.0 (Hugging Face tag).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open text-to-speech model from 2Noise with published weights for non-commercial use.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/2Noise/ChatTTS", "https://github.com/2noise/ChatTTS" ], "unverified": [ "summary" ] }, { "id": "zyphra-zonos-v0-1", "name": "Zonos v0.1", "level": "model", "family_id": null, "developer": [ "zyphra" ], "model_class": "speech synthesis", "modalities_in": [ "text", "audio" ], "modalities_out": [ "audio" ], "architecture": "hybrid", "architecture_detail": null, "params_total": 1651820416, "params_active": null, "params_checkpoint": null, "params_note": "Hybrid (1.65B) and transformer (1.62B) versions.", "context_length": null, "release_date": "2025-02-06", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "Multilingual (200k hours of training speech)", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open-weight text-to-speech model with voice cloning from a few seconds of reference audio.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Zyphra/Zonos-v0.1-hybrid", "https://github.com/Zyphra/Zonos", "https://www.zyphra.com/post/beta-release-of-zonos-v0-1" ], "unverified": [] }, { "id": "maya-research-maya1", "name": "Maya1", "level": "model", "family_id": null, "developer": [ "maya-research" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 3300928512, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10-18", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open-weight text-to-speech model that designs the voice from a natural-language description, with inline emotion controls.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/maya-research/maya1" ], "unverified": [] }, { "id": "neuphonic-neutts-air", "name": "NeuTTS Air", "level": "model", "family_id": null, "developer": [ "neuphonic" ], "model_class": "speech synthesis", "modalities_in": [ "text", "audio" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": null, "params_total": 747930496, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-09-15", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "laptop", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small on-device text-to-speech model with voice cloning, paired with the NeuCodec audio codec.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/neuphonic/neutts-air" ], "unverified": [ "deployment_tiers" ] }, { "id": "openmoss-moss-tts-v1-5", "name": "MOSS-TTS v1.5", "level": "model", "family_id": null, "developer": [ "openmoss" ], "model_class": "speech synthesis", "modalities_in": [ "text", "audio" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": null, "params_total": 8489841664, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-05-25", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Continuation of MOSS-TTS 1.0 with zero-shot voice cloning, long-form speech and token-level duration control.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/OpenMOSS-Team/MOSS-TTS-v1.5" ], "unverified": [] }, { "id": "kyutai-tts-1-6b-en-fr", "name": "Kyutai TTS 1.6B (en, fr)", "level": "model", "family_id": null, "developer": [ "kyutai" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": null, "params_total": 1800000000, "params_active": null, "params_checkpoint": null, "params_note": "Actually 1.8B: a 1B backbone plus a 600M depth transformer.", "context_length": null, "release_date": "2025-06-30", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": "English, French", "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Streaming text-to-speech model based on delayed streams modelling.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/kyutai/tts-1.6b-en_fr", "https://github.com/kyutai-labs/delayed-streams-modeling", "https://arxiv.org/abs/2509.08753" ], "unverified": [] }, { "id": "microsoft-mai-voice-1", "name": "MAI-Voice-1", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "speech synthesis", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-08-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft AI speech generation model that keeps one speaker identity across long-form content; available in Microsoft Foundry.", "verified_on": "2026-10-12", "sources": [ "https://microsoft.ai/news/today-were-announcing-3-new-world-class-mai-models-available-in-foundry/", "https://microsoft.ai/news/two-new-in-house-models/" ], "unverified": [] }, { "id": "amphion-maskgct", "name": "MaskGCT", "level": "model", "family_id": null, "developer": [ "amphion" ], "model_class": "speech synthesis", "modalities_in": [ "text", "audio" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": "masked generative codec transformer, non-autoregressive", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-09-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Zero-shot non-autoregressive text-to-speech model that needs no explicit text-speech alignment.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2409.00750", "https://huggingface.co/amphion/MaskGCT", "https://github.com/open-mmlab/Amphion" ], "unverified": [] }, { "id": "amphion-vevo", "name": "Vevo", "level": "model", "family_id": null, "developer": [ "amphion" ], "model_class": "speech synthesis", "modalities_in": [ "audio", "text" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-02-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Zero-shot voice imitation model that controls timbre and speaking style with self-supervised disentanglement.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2502.07243", "https://huggingface.co/amphion/Vevo" ], "unverified": [] }, { "id": "amphion-vevo2", "name": "Vevo2", "level": "model", "family_id": null, "developer": [ "amphion" ], "model_class": "speech synthesis", "modalities_in": [ "audio", "text" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-08-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": "Weights under CC BY-NC-ND 4.0 (Hugging Face tag); no AURION licence record for it yet.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Unified controllable framework for speech and singing voice generation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2508.16332", "https://huggingface.co/amphion/Vevo2" ], "unverified": [] }, { "id": "amphion-metis", "name": "Metis", "level": "model", "family_id": null, "developer": [ "amphion" ], "model_class": "speech synthesis", "modalities_in": [ "audio", "text" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": "masked generative pre-training", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-02-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Foundation speech generation model fine-tuned for TTS, voice conversion, speaker extraction, enhancement and lip-to-speech.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2502.03128", "https://huggingface.co/amphion/Metis" ], "unverified": [] }, { "id": "mit-ast", "name": "Audio Spectrogram Transformer (AST)", "level": "model", "family_id": null, "developer": [ "mit" ], "model_class": "audio classification", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "dense transformer", "architecture_detail": "convolution-free attention over spectrogram patches", "params_total": 86594063, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-04-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "bsd-3-clause", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Purely attention-based audio classifier on spectrograms; the AudioSet fine-tuned checkpoint labels sound events.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2104.01778", "https://github.com/YuanGongND/ast", "https://huggingface.co/MIT/ast-finetuned-audioset-10-10-0.4593" ], "unverified": [] }, { "id": "google-yamnet", "name": "YAMNet", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "audio classification", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": "MobileNet v1", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU", "microcontroller" ], "footprint_note": "Ported to Qualcomm AI Hub, MediaPipe and the STM32 model zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Audio event classifier that predicts 521 AudioSet event classes from a 16 kHz waveform.", "verified_on": "2026-10-12", "sources": [ "https://github.com/tensorflow/tfhub.dev/blob/master/assets/docs/google/models/yamnet/1.md", "https://github.com/tensorflow/models/tree/master/research/audioset/yamnet", "https://www.kaggle.com/models/google/yamnet", "https://ai.google.dev/edge/mediapipe/solutions/audio/audio_classifier" ], "unverified": [ "licence" ] }, { "id": "google-vggish", "name": "VGGish", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "audio classification", "modalities_in": [ "audio" ], "modalities_out": null, "architecture": "CNN", "architecture_detail": "VGG-style network", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2016-09-29", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Audio embedding model trained on YouTube-8M that outputs 128-dimensional embeddings of sound content.", "verified_on": "2026-10-12", "sources": [ "https://github.com/tensorflow/tfhub.dev/blob/master/assets/docs/google/models/vggish/1.md", "https://arxiv.org/abs/1609.09430", "https://www.kaggle.com/models/google/vggish" ], "unverified": [ "licence" ] }, { "id": "google-perch", "name": "Perch (Bird Vocalization Classifier)", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "audio classification", "modalities_in": [ "audio" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": "EfficientNet-B1", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Perch 2.0 components are in the perch repository.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Global bird embedding and classification model trained on Xeno-Canto recordings covering more than 10,000 species.", "verified_on": "2026-10-12", "sources": [ "https://github.com/tensorflow/tfhub.dev/blob/master/assets/docs/google/models/bird-vocalization-classifier/1.md", "https://github.com/google-research/perch", "https://www.kaggle.com/models/google/bird-vocalization-classifier" ], "unverified": [] }, { "id": "google-surfperch", "name": "SurfPerch", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "audio classification", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-04-25", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Pretrained network for passive acoustic monitoring of coral reefs, trained on reef, bird and other sounds.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2404.16436", "https://github.com/google-research/perch", "https://www.kaggle.com/models/google/surfperch" ], "unverified": [] }, { "id": "google-humpback-whale-detector", "name": "Humpback whale detector", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "audio classification", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": "ResNet-50 on PCEN spectrograms", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "A multispecies whale classifier is also published.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Detector by NOAA and Google that flags humpback whale sounds in underwater passive acoustic recordings.", "verified_on": "2026-10-12", "sources": [ "https://github.com/tensorflow/tfhub.dev/blob/master/assets/docs/google/models/humpback_whale/1.md", "https://www.kaggle.com/models/google/humpback-whale", "https://www.kaggle.com/models/google/multispecies-whale" ], "unverified": [] }, { "id": "google-spice", "name": "SPICE", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "music and audio", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-10-25", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Self-supervised pitch estimation model that recognises the dominant pitch in sung audio; mobile-compatible.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1910.11664", "https://github.com/tensorflow/tfhub.dev/blob/master/assets/docs/google/models/spice/2.md", "https://www.kaggle.com/models/google/spice" ], "unverified": [] }, { "id": "birdnet-team-birdnet", "name": "BirdNET", "level": "model", "family_id": null, "developer": [ "birdnet-team" ], "model_class": "audio classification", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-sa-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": "Code MIT; models CC BY-NC-SA 4.0 (README licence section).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Bird sound identification models and analysis scripts for processing large amounts of audio in acoustic surveys.", "verified_on": "2026-10-12", "sources": [ "https://github.com/kahst/BirdNET-Analyzer", "https://www.kaggle.com/models/shadiakiki1/birdnet-analyzer" ], "unverified": [] }, { "id": "universitat-pompeu-fabra-maest", "name": "MAEST", "level": "model", "family_id": null, "developer": [ "universitat-pompeu-fabra" ], "model_class": "audio classification", "modalities_in": [ "audio" ], "modalities_out": [], "architecture": "dense transformer", "architecture_detail": null, "params_total": 86858128, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-09-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-sa-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Music audio transformers trained on Discogs labels for music representation and style tagging; also in Essentia.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2309.16418", "https://huggingface.co/mtg-upf/discogs-maest-30s-pw-129e" ], "unverified": [] }, { "id": "xiaomi-dasheng", "name": "Dasheng", "level": "model", "family_id": null, "developer": [ "xiaomi" ], "model_class": "audio classification", "modalities_in": [ "audio" ], "modalities_out": null, "architecture": "dense transformer", "architecture_detail": "masked audio encoder", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Base (85M) to 1.2B.", "context_length": null, "release_date": "2024-06-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "General-purpose self-supervised audio encoder for speech, environmental sound and music classification.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2406.06992", "https://github.com/XiaoMi/dasheng", "https://huggingface.co/mispeech/dasheng-base" ], "unverified": [] }, { "id": "meta-audiobox-aesthetics", "name": "Audiobox Aesthetics", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "audio classification", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": 103859112, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-02-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Automatic quality assessment model that scores the aesthetics of speech, music and sound.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2502.05139", "https://github.com/facebookresearch/audiobox-aesthetics", "https://huggingface.co/facebook/audiobox-aesthetics" ], "unverified": [] }, { "id": "tencent-muq", "name": "MuQ", "level": "model", "family_id": null, "developer": [ "tencent" ], "model_class": "music and audio", "modalities_in": [ "audio" ], "modalities_out": null, "architecture": "dense transformer", "architecture_detail": null, "params_total": 333440416, "params_active": null, "params_checkpoint": null, "params_note": "MuQ-MuLan is a music-text joint embedding model built on it.", "context_length": null, "release_date": "2025-01-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Self-supervised music representation model using Mel residual vector quantisation, for tagging and other music understanding tasks.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2501.01108", "https://github.com/tencent-ailab/MuQ", "https://huggingface.co/OpenMuQ/MuQ-large-msd-iter" ], "unverified": [] }, { "id": "m-a-p-mert", "name": "MERT", "level": "model", "family_id": null, "developer": [ "m-a-p" ], "model_class": "music and audio", "modalities_in": [ "audio" ], "modalities_out": null, "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "95M and 330M checkpoints.", "context_length": null, "release_date": "2023-05-31", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Self-supervised acoustic music understanding model for music tagging and analysis tasks.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2306.00107", "https://huggingface.co/m-a-p/MERT-v1-330M" ], "unverified": [] }, { "id": "google-soundstream", "name": "SoundStream", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "music and audio", "modalities_in": [ "audio" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": "convolutional encoder-decoder with residual vector quantiser", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-07-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Neural audio codec for speech, music and general audio; a SoundStream mel decoder for music is published.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2107.03312", "https://github.com/tensorflow/tfhub.dev/blob/master/assets/docs/google/models/soundstream/mel/decoder/music/1.md", "https://www.kaggle.com/models/google/soundstream" ], "unverified": [] }, { "id": "kyutai-mimi", "name": "Mimi", "level": "model", "family_id": null, "developer": [ "kyutai" ], "model_class": "music and audio", "modalities_in": [ "audio" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": "streaming neural codec, 12.5 Hz, 1.1 kbps", "params_total": 96151393, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-09-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Streaming neural audio codec that combines semantic and acoustic information into tokens; used by Moshi.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/kyutai/mimi", "https://arxiv.org/abs/2410.00037" ], "unverified": [] }, { "id": "meta-magnet", "name": "MAGNeT", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": "masked non-autoregressive transformer over EnCodec tokens", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Small and medium checkpoints for music and sound.", "context_length": null, "release_date": "2024-01-09", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Text-to-music and text-to-sound model from AudioCraft that generates audio tokens with a single non-autoregressive transformer.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2401.04577", "https://huggingface.co/facebook/magnet-medium-10secs", "https://github.com/facebookresearch/audiocraft" ], "unverified": [] }, { "id": "meta-sam-audio", "name": "SAM Audio", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "music and audio", "modalities_in": [ "audio", "text", "video" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-12-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "sam-license", "access_routes": [ "open weights download", "gated preview" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Segment Anything model for audio that isolates sounds in a mixture using text, visual or time-span prompts.", "verified_on": "2026-10-12", "sources": [ "https://github.com/facebookresearch/sam-audio", "https://huggingface.co/facebook/sam-audio-large", "https://ai.meta.com/blog/sam-audio/" ], "unverified": [] }, { "id": "funaudiollm-inspiremusic", "name": "InspireMusic", "level": "model", "family_id": null, "developer": [ "funaudiollm" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "hybrid", "architecture_detail": "autoregressive transformer with super-resolution flow matching", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-02-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Long-form music, song and audio generation framework combining a language model with super-resolution flow matching.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2503.00084", "https://github.com/FunAudioLLM/InspireMusic", "https://www.modelscope.cn/models/iic/InspireMusic-Base-24kHz" ], "unverified": [ "licence" ] }, { "id": "funaudiollm-thinksound", "name": "ThinkSound", "level": "model", "family_id": null, "developer": [ "funaudiollm" ], "model_class": "music and audio", "modalities_in": [ "video", "text" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-06-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Video-to-audio generation and editing that uses chain-of-thought reasoning in a multimodal LLM.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2506.21448", "https://huggingface.co/FunAudioLLM/ThinkSound" ], "unverified": [] }, { "id": "mmaudio", "name": "MMAudio", "level": "model", "family_id": null, "developer": [ "uiuc", "sony" ], "model_class": "music and audio", "modalities_in": [ "video", "text" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-12-19", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Video-to-audio synthesis model by Ho Kei Cheng and co-authors, trained jointly on video and text-audio data.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2412.15322", "https://huggingface.co/hkchengrex/MMAudio", "https://arxiv.org/html/2412.15322" ], "unverified": [ "developer" ] }, { "id": "foleycrafter", "name": "FoleyCrafter", "level": "model", "family_id": null, "developer": [ "shanghai-ai-lab" ], "model_class": "music and audio", "modalities_in": [ "video", "text" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-07-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Generates sound effects synchronised with silent video.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2407.01494", "https://github.com/open-mmlab/FoleyCrafter", "https://arxiv.org/html/2407.01494" ], "unverified": [ "developer" ] }, { "id": "google-ddsp", "name": "DDSP", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "music and audio", "modalities_in": [ "audio" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": "differentiable signal processing modules", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-01-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Library and models that combine classic signal processing with deep learning for audio and timbre synthesis.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2001.04643", "https://github.com/magenta/ddsp" ], "unverified": [] }, { "id": "inria-open-unmix", "name": "Open-Unmix", "level": "model", "family_id": null, "developer": [ "inria" ], "model_class": "music and audio", "modalities_in": [ "audio" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Pre-trained models include umxl (2021) and a speech enhancement model umxse from Sony.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Reference deep network for music source separation, splitting songs into stems.", "verified_on": "2026-10-12", "sources": [ "https://github.com/sigsep/open-unmix-pytorch", "https://pytorch.org/hub/sigsep_open-unmix-pytorch_umx" ], "unverified": [] }, { "id": "alibaba-frcrn", "name": "FRCRN", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "music and audio", "modalities_in": [ "audio" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": "convolutional recurrent network with frequency recurrence", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-06-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Monaural speech enhancement model, available in ClearerVoice-Studio and on ModelScope.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2206.07293", "https://github.com/modelscope/ClearerVoice-Studio" ], "unverified": [ "developer" ] }, { "id": "alibaba-mossformer2", "name": "MossFormer2", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "music and audio", "modalities_in": [ "audio" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-12-19", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Monaural speech separation model combining a transformer with an RNN-free recurrent network.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2312.11825", "https://github.com/modelscope/ClearerVoice-Studio" ], "unverified": [ "developer" ] }, { "id": "alibaba-zipenhancer", "name": "ZipEnhancer", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "music and audio", "modalities_in": [ "audio" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-01-09", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Monaural speech enhancement model with a dual-path down-up sampling Zipformer.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2501.05183", "https://www.modelscope.cn/models/iic/speech_zipenhancer_ans_multiloss_16k_base" ], "unverified": [ "developer" ] }, { "id": "riffusion", "name": "Riffusion", "level": "model", "family_id": null, "developer": [ "riffusion" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "diffusion", "architecture_detail": "Stable Diffusion fine-tuned on spectrogram images", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-12-13", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": "CreativeML OpenRAIL-M (Hugging Face tag); no AURION licence record for it yet.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time music generation with Stable Diffusion applied to spectrograms.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/riffusion/riffusion-model-v1", "https://www.riffusion.com/about" ], "unverified": [] }, { "id": "heartmula-heartmula-oss-3b", "name": "HeartMuLa-oss-3B", "level": "model", "family_id": null, "developer": [ "heartmula" ], "model_class": "music and audio", "modalities_in": [ "text" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": null, "params_total": 3938113536, "params_active": null, "params_checkpoint": null, "params_note": "Paired with the HeartCodec-oss codec.", "context_length": null, "release_date": "2026-01-14", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open music generation language model released with its own audio codec.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/HeartMuLa/HeartMuLa-oss-3B" ], "unverified": [ "summary" ] }, { "id": "tabnet", "name": "TabNet", "level": "family", "family_id": null, "developer": [], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "dense transformer", "architecture_detail": "Sequential attention that selects features at each decision step", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-08-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the dreamquark-ai/tabnet PyTorch implementation (MIT); the paper itself carries no licence.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Deep learning model for tabular data that uses sequential attention to choose which features to reason from at each step, which makes its decisions interpretable. Paper by Arik and Pfister (2019).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1908.07442", "https://github.com/dreamquark-ai/tabnet", "https://github.com/manujosephv/pytorch_tabular", "https://modelzoo.co/model/tabnet" ], "unverified": [] }, { "id": "node-tabular", "name": "NODE (Neural Oblivious Decision Ensembles)", "level": "family", "family_id": null, "developer": [ "yandex" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": "Differentiable ensembles of oblivious decision trees", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-09-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the authors' Qwicen/node implementation (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Deep learning architecture for tabular data built from differentiable oblivious decision tree ensembles. Paper by Popov et al. (2019).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1909.06312", "https://github.com/Qwicen/node", "https://github.com/manujosephv/pytorch_tabular", "https://arxiv.org/html/1909.06312" ], "unverified": [] }, { "id": "yandex-ft-transformer", "name": "FT-Transformer", "level": "family", "family_id": null, "developer": [ "yandex" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "dense transformer", "architecture_detail": "Feature tokenizer plus transformer; the same paper also defines the ResNet-like tabular baseline", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-06-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the yandex-research/rtdl-revisiting-models repository (Apache-2.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Transformer for tabular data that turns each feature into a token, from the paper Revisiting Deep Learning Models for Tabular Data (Gorishniy et al., 2021). A standard deep tabular baseline.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2106.11959", "https://github.com/yandex-research/rtdl-revisiting-models", "https://github.com/manujosephv/pytorch_tabular" ], "unverified": [] }, { "id": "tabtransformer", "name": "TabTransformer", "level": "family", "family_id": null, "developer": [], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "dense transformer", "architecture_detail": "Transformer layers produce contextual embeddings of categorical features", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-12-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the pytorch-tabular implementation (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tabular model that passes categorical feature embeddings through transformer layers to make them contextual. Paper by Huang et al. (2020).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2012.06678", "https://github.com/pytorch-tabular/pytorch_tabular", "https://github.com/manujosephv/pytorch_tabular" ], "unverified": [] }, { "id": "autoint", "name": "AutoInt", "level": "family", "family_id": null, "developer": [ "peking-university" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "dense transformer", "architecture_detail": "Multi-head self-attention over feature embeddings", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-10-29", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the pytorch-tabular implementation (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Model that learns feature interactions automatically with self-attention, first proposed for click-through rate prediction and used as a tabular baseline. Paper by Song et al. (2018).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1810.11921", "https://github.com/pytorch-tabular/pytorch_tabular", "https://github.com/manujosephv/pytorch_tabular", "https://github.com/RUCAIBox/RecBole2.0/blob/main/model_list.md", "https://arxiv.org/html/1810.11921" ], "unverified": [] }, { "id": "gandalf", "name": "GANDALF", "level": "family", "family_id": null, "developer": [], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": "Gated feature learning units", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-07-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the pytorch-tabular implementation (MIT), maintained by the paper's first author. GATE (Gated Additive Tree Ensemble) is described in earlier versions of the same arXiv paper.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Gated Adaptive Network for Deep Automated Learning of Features, a deep tabular model built from gated feature learning units. Paper by Joseph and Raj (2022).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2207.08548", "https://github.com/pytorch-tabular/pytorch_tabular", "https://github.com/manujosephv/pytorch_tabular", "https://arxiv.org/html/2207.08548" ], "unverified": [] }, { "id": "danets", "name": "DANets (Deep Abstract Networks)", "level": "family", "family_id": null, "developer": [], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-12-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the pytorch-tabular implementation (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Deep Abstract Networks for tabular classification and regression. Paper by Chen et al. (2021).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2112.02962", "https://github.com/pytorch-tabular/pytorch_tabular", "https://github.com/manujosephv/pytorch_tabular" ], "unverified": [] }, { "id": "microsoft-ebm", "name": "Explainable Boosting Machine (EBM)", "level": "family", "family_id": null, "developer": [ "microsoft" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "gradient boosting", "architecture_detail": "Glass-box generalised additive model trained with boosting", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Part of the InterpretML library (interpretml/interpret, MIT; repository created 3 May 2019).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Interpretable tabular model from Microsoft's InterpretML library: a boosted generalised additive model whose per-feature contributions can be read directly.", "verified_on": "2026-10-12", "sources": [ "https://github.com/interpretml/interpret", "https://auto.gluon.ai/stable/api/autogluon.tabular.models.html" ], "unverified": [] }, { "id": "google-tapas", "name": "TAPAS", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "tabular", "modalities_in": [ "table", "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "BERT-style encoder pre-trained jointly on text and tables", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-04-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google model that answers questions over tables without generating logical forms, pre-trained on text and tables. Paper by Herzig et al. (2020).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2004.02349", "https://github.com/google-research/tapas", "https://huggingface.co/google/tapas-base-finetuned-wtq", "https://www.modelscope.cn/models/google/tapas-base-finetuned-wtq" ], "unverified": [] }, { "id": "google-tapas-base-finetuned-wtq", "name": "TAPAS base fine-tuned on WTQ", "level": "variant", "family_id": "google-tapas", "developer": [ "google" ], "model_class": "tabular", "modalities_in": [ "table", "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "A large version (google/tapas-large-finetuned-wtq) and other sizes and tasks are also published.", "context_length": null, "release_date": "2022-03-02", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "TAPAS base checkpoint fine-tuned in a chain on SQA, WikiSQL and WikiTable Questions for table question answering.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/google/tapas-base-finetuned-wtq", "https://huggingface.co/api/models?pipeline_tag=table-question-answering&sort=downloads&limit=50" ], "unverified": [] }, { "id": "microsoft-tapex", "name": "TAPEX", "level": "family", "family_id": null, "developer": [ "microsoft" ], "model_class": "tabular", "modalities_in": [ "table", "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "Pre-trained by learning to act as a neural SQL executor", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-07-16", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft table pre-training approach that teaches a language model to execute SQL queries, giving it table reasoning skills for table question answering. Paper by Liu et al. (2021).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2107.07653", "https://github.com/microsoft/Table-Pretraining", "https://huggingface.co/microsoft/tapex-base-finetuned-wikisql", "https://www.modelscope.cn/models/microsoft/tapex-large" ], "unverified": [] }, { "id": "microsoft-tapex-base-finetuned-wikisql", "name": "TAPEX base fine-tuned on WikiSQL", "level": "variant", "family_id": "microsoft-tapex", "developer": [ "microsoft" ], "model_class": "tabular", "modalities_in": [ "table", "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "tapex-large and other fine-tunes are also published.", "context_length": null, "release_date": "2022-03-02", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Base-size TAPEX checkpoint fine-tuned on WikiSQL for table question answering.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/microsoft/tapex-base-finetuned-wikisql", "https://huggingface.co/api/models?pipeline_tag=table-question-answering&sort=downloads&limit=50" ], "unverified": [] }, { "id": "tabstar", "name": "TabSTAR", "level": "family", "family_id": null, "developer": [ "technion" ], "model_class": "tabular", "modalities_in": [ "table", "text" ], "modalities_out": [ "table" ], "architecture": "dense transformer", "architecture_detail": "Fine-tuned from intfloat/e5-small-v2 (Hugging Face base_model)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-05-23", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tabular foundation model for tables that contain free-text fields, used through the tabstar Python package. Paper by Arazi et al. (2025).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2505.18125", "https://huggingface.co/alana89/TabSTAR", "https://huggingface.co/api/models?pipeline_tag=tabular-classification&sort=downloads&limit=50", "https://arxiv.org/html/2505.18125" ], "unverified": [] }, { "id": "synthefy-nori", "name": "Nori", "level": "family", "family_id": null, "developer": [ "synthefy" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "dense transformer", "architecture_detail": "In-context learning on labelled rows; trained entirely on synthetic data", "params_total": 5900000, "params_active": null, "params_checkpoint": null, "params_note": "About 5.9M parameters (model card). The coverage file also lists 30M and 100M sizes.", "context_length": null, "release_date": "2026-05-24", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Synthefy tabular foundation model for regression: given a few labelled rows as context it predicts new rows in one forward pass, with no task-specific training.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Synthefy/Nori", "https://github.com/Synthefy/synthefy-nori", "https://auto.gluon.ai/stable/api/autogluon.tabular.models.html" ], "unverified": [ "params_note" ] }, { "id": "google-tabfm-1-0", "name": "TabFM 1.0", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "dense transformer", "architecture_detail": "Zero-shot in-context model; PyTorch and JAX weights", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-06-29", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": "Weights under the TabFM Non-Commercial License v1.0 (model card).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Zero-shot tabular foundation model from Google Research for classification and regression on mixed numerical and categorical columns; training rows are passed as context.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/google/tabfm-1.0.0-pytorch", "https://huggingface.co/api/models?pipeline_tag=tabular-classification&sort=downloads&limit=50" ], "unverified": [] }, { "id": "nrel-routee-powertrain-model-library", "name": "RouteE-Powertrain model library", "level": "family", "family_id": null, "developer": [], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "282 models covering 74 vehicle configurations across 25 makes (model card).", "context_length": null, "release_date": "2026-07-28", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "bsd-3-clause", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Library of pre-trained vehicle energy models that predict how much energy a given vehicle uses on a road link from speed, grade and turn angle; used by routing engines to compute energy-aware routes.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/NatLabRockies/routee-powertrain-model-library", "https://github.com/NatLabRockies/routee-powertrain", "https://huggingface.co/api/models?pipeline_tag=tabular-regression&sort=downloads&limit=50" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "pyod", "name": "PyOD", "level": "family", "family_id": null, "developer": [ "carnegie-mellon-university" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": "Python library of outlier detectors", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-01-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": "Repository licence BSD-2-Clause (GitHub API); no AURION licence record exists for it yet.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Python library of anomaly and outlier detection algorithms (60 or more detectors per its repository) for tabular, time series, graph and other data. Paper by Zhao et al. (2019).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1901.01588", "https://github.com/yzhao062/pyod", "https://pyod.readthedocs.io/en/latest/", "https://arxiv.org/html/1901.01588" ], "unverified": [] }, { "id": "ecod", "name": "ECOD", "level": "family", "family_id": null, "developer": [ "northeastern-university", "carnegie-mellon-university" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": "Unsupervised, parameter-free; uses empirical cumulative distribution functions", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-01-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": "Implemented in PyOD (BSD-2-Clause).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Unsupervised outlier detector that scores points by how extreme they are under each feature's empirical cumulative distribution. Paper by Li et al. (2022).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2201.00382", "https://github.com/yzhao062/pyod", "https://pyod.readthedocs.io/en/latest/", "https://arxiv.org/html/2201.00382" ], "unverified": [] }, { "id": "copod", "name": "COPOD", "level": "family", "family_id": null, "developer": [ "northeastern-university", "carnegie-mellon-university" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": "Copula-based outlier detection", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-09-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": "Implemented in PyOD (BSD-2-Clause).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Copula-based outlier detector that estimates tail probabilities for each data point. Paper by Li et al. (2020).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2009.09463", "https://github.com/yzhao062/pyod", "https://pyod.readthedocs.io/en/latest/", "https://arxiv.org/html/2009.09463" ], "unverified": [] }, { "id": "deep-isolation-forest", "name": "Deep Isolation Forest (DIF)", "level": "family", "family_id": null, "developer": [ "nudt" ], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": "Isolation forest on random neural representations", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-06-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": "Implemented in PyOD (BSD-2-Clause).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Anomaly detector that applies isolation forests to data mapped through randomly initialised neural networks. Paper by Xu et al. (2022).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2206.06602", "https://github.com/yzhao062/pyod", "https://pyod.readthedocs.io/en/latest/", "https://arxiv.org/html/2206.06602" ], "unverified": [] }, { "id": "local-outlier-factor", "name": "LOF (Local Outlier Factor)", "level": "family", "family_id": null, "developer": [], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": "Density-based outlier score relative to neighbours", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": "Implemented in PyOD (BSD-2-Clause). Reference paper date not stated on the pages read.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Classical density-based outlier detector that compares the local density of each point with that of its neighbours. Listed among PyOD's implemented algorithms.", "verified_on": "2026-10-12", "sources": [ "https://pyod.readthedocs.io/en/latest/", "https://github.com/yzhao062/pyod" ], "unverified": [] }, { "id": "one-class-svm", "name": "OCSVM (One-Class SVM)", "level": "family", "family_id": null, "developer": [], "model_class": "tabular", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": "Support vector method fitted to normal data only", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": "Implemented in PyOD (BSD-2-Clause). Reference paper date not stated on the pages read.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Classical novelty and outlier detector that learns a boundary around normal data with a support vector machine. Listed among PyOD's implemented algorithms.", "verified_on": "2026-10-12", "sources": [ "https://pyod.readthedocs.io/en/latest/", "https://github.com/yzhao062/pyod" ], "unverified": [] }, { "id": "n-beats", "name": "N-BEATS", "level": "family", "family_id": null, "developer": [ "element-ai", "mila" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": "Deep stack of fully connected blocks with backward and forward residual links", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-05-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the Nixtla neuralforecast implementation (Apache-2.0). NeuralForecast also implements N-BEATSx with exogenous variables.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Neural basis expansion model for univariate point forecasting with an interpretable configuration. Paper by Oreshkin et al. (2019).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1905.10437", "https://github.com/Nixtla/neuralforecast", "https://nixtlaverse.nixtla.io/neuralforecast/docs/capabilities/overview.html", "https://arxiv.org/html/1905.10437" ], "unverified": [] }, { "id": "n-hits", "name": "N-HiTS", "level": "family", "family_id": null, "developer": [], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": "Hierarchical interpolation and multi-rate sampling on top of N-BEATS-style blocks", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-01-30", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the Nixtla neuralforecast implementation (Apache-2.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Neural hierarchical interpolation model for long-horizon forecasting. Paper by Challu et al. (2022).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2201.12886", "https://github.com/Nixtla/neuralforecast", "https://nixtlaverse.nixtla.io/neuralforecast/docs/capabilities/overview.html", "https://www.kaggle.com/models/kp-labs/clean-nhits-model" ], "unverified": [] }, { "id": "temporal-fusion-transformer", "name": "Temporal Fusion Transformer (TFT)", "level": "family", "family_id": null, "developer": [ "university-of-oxford", "google" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "hybrid", "architecture_detail": "Recurrent layers for local processing plus interpretable self-attention", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-12-19", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the Nixtla neuralforecast implementation (Apache-2.0). NVIDIA Deep Learning Examples publishes trained checkpoints (electricity, traffic) on NGC.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Attention-based model for interpretable multi-horizon forecasting with static, known and observed inputs. Paper by Lim et al. (2019).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1912.09363", "https://github.com/Nixtla/neuralforecast", "https://nixtlaverse.nixtla.io/neuralforecast/docs/capabilities/overview.html", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/dle/models/tft_base_pyt_ckpt_ds-electricity", "https://arxiv.org/html/1912.09363" ], "unverified": [] }, { "id": "intel-time-series-forecasting-electricity-0001", "name": "Intel time-series-forecasting-electricity-0001", "level": "variant", "family_id": "temporal-fusion-transformer", "developer": [ "intel" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "hybrid", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Temporal Fusion Transformer model for electricity consumption forecasting, published in the OpenVINO Open Model Zoo.", "verified_on": "2026-10-12", "sources": [ "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/intel/time-series-forecasting-electricity-0001" ], "unverified": [] }, { "id": "patchtst", "name": "PatchTST", "level": "family", "family_id": null, "developer": [ "princeton-university", "ibm" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "dense transformer", "architecture_detail": "Channel-independent transformer on patches of the series", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-11-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the official yuqinie98/PatchTST repository (Apache-2.0). Not the same as IBM PatchTST-FM, which AURION holds separately.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Transformer for long-term forecasting that splits each series into patches treated as tokens, from the paper A Time Series is Worth 64 Words (Nie et al., 2022).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2211.14730", "https://github.com/yuqinie98/PatchTST", "https://nixtlaverse.nixtla.io/neuralforecast/docs/capabilities/overview.html", "https://arxiv.org/html/2211.14730" ], "unverified": [] }, { "id": "deepar", "name": "DeepAR", "level": "family", "family_id": null, "developer": [ "amazon" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "autoregressive", "architecture_detail": "Autoregressive recurrent network trained on many related series", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-04-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the GluonTS and neuralforecast implementations (Apache-2.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Probabilistic forecasting method that trains an autoregressive recurrent network across many related time series. Paper by Salinas et al. (2017).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1704.04110", "https://github.com/awslabs/gluonts", "https://github.com/Nixtla/neuralforecast", "https://nixtlaverse.nixtla.io/neuralforecast/docs/capabilities/overview.html", "https://arxiv.org/html/1704.04110" ], "unverified": [] }, { "id": "amazon-mq-rnn", "name": "MQ-RNN and MQ-CNN", "level": "family", "family_id": null, "developer": [ "amazon" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": "Sequence-to-sequence quantile forecaster with recurrent or convolutional encoder", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-11-29", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the GluonTS implementation (Apache-2.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multi-horizon quantile recurrent forecaster, demonstrated on forecasting demand for items sold on Amazon. Paper by Wen et al. (2017).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1711.11053", "https://github.com/awslabs/gluonts" ], "unverified": [] }, { "id": "informer", "name": "Informer", "level": "family", "family_id": null, "developer": [], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "dense transformer", "architecture_detail": "ProbSparse self-attention for long sequences", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-12-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the official zhouhaoyi/Informer2020 repository (Apache-2.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Efficient transformer for long sequence time series forecasting (AAAI 2021). Paper by Zhou et al. (2020).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2012.07436", "https://github.com/zhouhaoyi/Informer2020", "https://nixtlaverse.nixtla.io/neuralforecast/docs/capabilities/overview.html" ], "unverified": [] }, { "id": "autoformer", "name": "Autoformer", "level": "family", "family_id": null, "developer": [ "tsinghua-university" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "dense transformer", "architecture_detail": "Series decomposition with an auto-correlation mechanism", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-06-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of thuml/Time-Series-Library (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Decomposition transformer with auto-correlation for long-term forecasting. Paper by Wu et al. (2021).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2106.13008", "https://github.com/thuml/Time-Series-Library", "https://nixtlaverse.nixtla.io/neuralforecast/docs/capabilities/overview.html" ], "unverified": [] }, { "id": "fedformer", "name": "FEDformer", "level": "family", "family_id": null, "developer": [ "alibaba-group" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "dense transformer", "architecture_detail": "Frequency-enhanced decomposed transformer", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-01-30", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of thuml/Time-Series-Library (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Frequency enhanced decomposed transformer for long-term forecasting. Paper by Zhou et al. (2022).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2201.12740", "https://github.com/thuml/Time-Series-Library", "https://nixtlaverse.nixtla.io/neuralforecast/docs/capabilities/overview.html", "https://arxiv.org/html/2201.12740" ], "unverified": [] }, { "id": "itransformer", "name": "iTransformer", "level": "family", "family_id": null, "developer": [ "tsinghua-university" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "dense transformer", "architecture_detail": "Attention across variates instead of across time steps", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-10-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the official thuml/iTransformer repository (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Inverted transformer that embeds each variate as a token, for multivariate forecasting (ICLR 2024). Paper by Liu et al. (2023).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2310.06625", "https://github.com/thuml/iTransformer", "https://nixtlaverse.nixtla.io/neuralforecast/docs/capabilities/overview.html" ], "unverified": [] }, { "id": "timesnet", "name": "TimesNet", "level": "family", "family_id": null, "developer": [ "tsinghua-university" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "CNN", "architecture_detail": "Reshapes 1D series into 2D by period and applies 2D convolution", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-10-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of thuml/Time-Series-Library (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "General time series model for forecasting, imputation, classification and anomaly detection that models variation within and across periods in 2D. Paper by Wu et al. (2022).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2210.02186", "https://github.com/thuml/Time-Series-Library", "https://nixtlaverse.nixtla.io/neuralforecast/docs/capabilities/overview.html" ], "unverified": [] }, { "id": "tide", "name": "TiDE", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": "MLP-based encoder and decoder", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-04-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the Nixtla neuralforecast implementation (Apache-2.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Time-series Dense Encoder, an MLP model for long-term forecasting with covariates. Paper by Das et al. (2023).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2304.08424", "https://github.com/Nixtla/neuralforecast", "https://nixtlaverse.nixtla.io/neuralforecast/docs/capabilities/overview.html", "https://arxiv.org/html/2304.08424" ], "unverified": [] }, { "id": "tsmixer", "name": "TSMixer", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": "All-MLP mixing across time and features", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-03-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the Nixtla neuralforecast implementation (Apache-2.0), which also has TSMixerx with exogenous inputs.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "All-MLP architecture for time series forecasting that mixes along time and feature dimensions. Paper by Chen et al. (2023).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2303.06053", "https://github.com/Nixtla/neuralforecast", "https://nixtlaverse.nixtla.io/neuralforecast/docs/capabilities/overview.html", "https://arxiv.org/html/2303.06053" ], "unverified": [] }, { "id": "timemixer", "name": "TimeMixer", "level": "family", "family_id": null, "developer": [ "ant-group", "tsinghua-university" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": "Decomposable multiscale mixing", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-05-23", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of thuml/Time-Series-Library (MIT). Not the same as IBM TinyTimeMixer (TTM).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Forecasting model that mixes seasonal and trend components across several time scales. Paper by Wang et al. (2024).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2405.14616", "https://github.com/thuml/Time-Series-Library", "https://nixtlaverse.nixtla.io/neuralforecast/docs/capabilities/overview.html", "https://arxiv.org/html/2405.14616" ], "unverified": [] }, { "id": "dlinear", "name": "DLinear and NLinear", "level": "family", "family_id": null, "developer": [ "cuhk", "idea-research" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": "One-layer linear models, with trend and seasonal decomposition (DLinear) or last-value normalisation (NLinear)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-05-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the official cure-lab/LTSF-Linear repository (Apache-2.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Simple linear forecasting baselines from the paper Are Transformers Effective for Time Series Forecasting? (Zeng et al., 2022), which matched or beat transformer models on long-term benchmarks.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2205.13504", "https://github.com/cure-lab/LTSF-Linear", "https://nixtlaverse.nixtla.io/neuralforecast/docs/capabilities/overview.html", "https://arxiv.org/html/2205.13504" ], "unverified": [] }, { "id": "time-llm", "name": "Time-LLM", "level": "family", "family_id": null, "developer": [ "monash-university", "ant-group" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "dense transformer", "architecture_detail": "Reprograms a frozen large language model for forecasting", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-10-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the official KimMeen/Time-LLM repository (Apache-2.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Method that reprograms a frozen large language model to forecast time series (ICLR 2024). Paper by Jin et al. (2023).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2310.01728", "https://github.com/KimMeen/Time-LLM", "https://nixtlaverse.nixtla.io/neuralforecast/docs/capabilities/overview.html", "https://arxiv.org/html/2310.01728" ], "unverified": [] }, { "id": "tcn", "name": "TCN (Temporal Convolutional Network)", "level": "family", "family_id": null, "developer": [ "carnegie-mellon-university", "intel" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "CNN", "architecture_detail": "Causal dilated convolutions with residual blocks", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-03-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the authors' locuslab/TCN repository (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Generic convolutional architecture for sequence modelling, shown to match or beat recurrent networks on many sequence tasks. Paper by Bai et al. (2018).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1803.01271", "https://github.com/locuslab/TCN", "https://nixtlaverse.nixtla.io/neuralforecast/docs/capabilities/overview.html", "https://modelzoo.co/model/tcn-pytorch", "https://huggingface.co/datasets/pwc-archive/methods", "https://arxiv.org/html/1803.01271" ], "unverified": [] }, { "id": "lstnet", "name": "LSTNet", "level": "family", "family_id": null, "developer": [ "carnegie-mellon-university" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "hybrid", "architecture_detail": "Convolutional and recurrent layers with a skip-recurrent component and an autoregressive part", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-03-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the GluonTS implementation (Apache-2.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Long- and short-term time series network for multivariate forecasting. Paper by Lai et al. (2017).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1703.07015", "https://github.com/awslabs/gluonts", "https://github.com/onnx/models/blob/main/README.md", "https://arxiv.org/html/1703.07015" ], "unverified": [] }, { "id": "rocket", "name": "ROCKET", "level": "family", "family_id": null, "developer": [ "monash-university" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "CNN", "architecture_detail": "Random convolutional kernels feeding a linear classifier", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-10-29", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "gpl-3-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the authors' angus924/rocket repository (GPL-3.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fast time series classification method that transforms series with many random convolutional kernels and trains a linear classifier. Paper by Dempster et al. (2019).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1910.13051", "https://github.com/angus924/rocket", "https://huggingface.co/datasets/pwc-archive/methods", "https://arxiv.org/html/1910.13051" ], "unverified": [] }, { "id": "inceptiontime", "name": "InceptionTime", "level": "family", "family_id": null, "developer": [ "universite-de-haute-alsace" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "CNN", "architecture_detail": "Ensemble of Inception-style convolutional networks", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-09-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "gpl-3-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the authors' hfawaz/InceptionTime repository (GPL-3.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Deep convolutional ensemble for time series classification. Paper by Ismail Fawaz et al. (2019).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1909.04939", "https://github.com/hfawaz/InceptionTime", "https://huggingface.co/datasets/pwc-archive/methods", "https://arxiv.org/html/1909.04939" ], "unverified": [] }, { "id": "mstl", "name": "MSTL", "level": "family", "family_id": null, "developer": [ "university-of-melbourne" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": "Seasonal-trend decomposition for multiple seasonal periods", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-07-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the Nixtla statsforecast implementation (Apache-2.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Seasonal-trend decomposition algorithm for series with several seasonal patterns, used as a forecasting model in statsforecast. Paper by Bandara et al. (2021).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2107.13462", "https://github.com/Nixtla/statsforecast", "https://nixtlaverse.nixtla.io/statsforecast/index.html", "https://arxiv.org/html/2107.13462" ], "unverified": [] }, { "id": "ets", "name": "ETS (exponential smoothing)", "level": "family", "family_id": null, "developer": [], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": "Error, trend and seasonal state space models; AutoETS selects the form automatically", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the Nixtla statsforecast implementation (Apache-2.0). Classical method; reference paper date not stated on the pages read.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Classical exponential smoothing forecaster in state space form, available as AutoETS in statsforecast. A standard statistical baseline.", "verified_on": "2026-10-12", "sources": [ "https://nixtlaverse.nixtla.io/statsforecast/index.html", "https://github.com/Nixtla/statsforecast" ], "unverified": [] }, { "id": "theta", "name": "Theta method", "level": "family", "family_id": null, "developer": [], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": "AutoTheta selects among standard, optimised and dynamic variants", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the Nixtla statsforecast implementation (Apache-2.0). Classical method; reference paper date not stated on the pages read.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Classical decomposition-based forecasting method, available as AutoTheta and related models in statsforecast. A standard statistical baseline.", "verified_on": "2026-10-12", "sources": [ "https://nixtlaverse.nixtla.io/statsforecast/index.html", "https://github.com/Nixtla/statsforecast" ], "unverified": [] }, { "id": "croston", "name": "Croston method", "level": "family", "family_id": null, "developer": [], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": "Classic, optimised and SBA variants", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the Nixtla statsforecast implementation (Apache-2.0). Classical method; reference paper date not stated on the pages read.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Classical forecaster for intermittent demand, where many periods have zero sales, available in statsforecast.", "verified_on": "2026-10-12", "sources": [ "https://nixtlaverse.nixtla.io/statsforecast/index.html", "https://github.com/Nixtla/statsforecast" ], "unverified": [] }, { "id": "ibm-granite-ttm-r3", "name": "Granite TimeSeries TTM r3", "level": "model", "family_id": null, "developer": [ "ibm" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "MoE", "architecture_detail": "Mixture of experts over tiny models of varying complexity (model card)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Experts of 1M to 35M parameters, Lite 1M to 18M (model card).", "context_length": null, "release_date": "2026-05-21", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Third generation of IBM's TinyTimeMixer forecasting models, for fast zero-shot and few-shot forecasting in production.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/ibm-granite/granite-timeseries-ttm-r3", "https://huggingface.co/api/models?pipeline_tag=time-series-forecasting&sort=downloads&limit=50" ], "unverified": [] }, { "id": "neoquasar-kronos", "name": "Kronos", "level": "family", "family_id": null, "developer": [], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "autoregressive", "architecture_detail": "Decoder-only transformer over tokenised K-line (candlestick) sequences", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "mini 4.1M, small 24.7M, base 102.3M; large 499.2M not publicly available (model card).", "context_length": null, "release_date": "2025-08-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Family of open foundation models for financial candlestick (K-line) data, trained on data from over 45 exchanges, for forecasting price series. Paper by Shi et al. (2025).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2508.02739", "https://huggingface.co/NeoQuasar/Kronos-base", "https://github.com/shiyu-coder/Kronos", "https://huggingface.co/api/models?pipeline_tag=time-series-forecasting&sort=downloads&limit=50" ], "unverified": [] }, { "id": "neoquasar-kronos-mini", "name": "Kronos-mini", "level": "variant", "family_id": "neoquasar-kronos", "developer": [], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "autoregressive", "architecture_detail": null, "params_total": 4100000, "params_active": null, "params_checkpoint": null, "params_note": "Context length is the maximum K-line sequence length (model card).", "context_length": 2048, "release_date": "2025-07-01", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Kronos mini checkpoint for financial candlestick forecasting.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/NeoQuasar/Kronos-mini", "https://huggingface.co/NeoQuasar/Kronos-base" ], "unverified": [] }, { "id": "neoquasar-kronos-small", "name": "Kronos-small", "level": "variant", "family_id": "neoquasar-kronos", "developer": [], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "autoregressive", "architecture_detail": null, "params_total": 24700000, "params_active": null, "params_checkpoint": null, "params_note": "Context length is the maximum K-line sequence length (model card).", "context_length": 512, "release_date": "2025-06-30", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Kronos small checkpoint for financial candlestick forecasting.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/NeoQuasar/Kronos-small", "https://huggingface.co/NeoQuasar/Kronos-base" ], "unverified": [] }, { "id": "neoquasar-kronos-base", "name": "Kronos-base", "level": "variant", "family_id": "neoquasar-kronos", "developer": [], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "autoregressive", "architecture_detail": null, "params_total": 102300000, "params_active": null, "params_checkpoint": null, "params_note": "Context length is the maximum K-line sequence length (model card).", "context_length": 512, "release_date": "2025-06-30", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Kronos base checkpoint for financial candlestick forecasting.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/NeoQuasar/Kronos-base", "https://huggingface.co/NeoQuasar/Kronos-base" ], "unverified": [] }, { "id": "chattime", "name": "ChatTime", "level": "family", "family_id": null, "developer": [], "model_class": "time series", "modalities_in": [ "time series", "text" ], "modalities_out": [ "time series", "text" ], "architecture": "autoregressive", "architecture_detail": "Treats time series as a foreign language for a language model", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-12-16", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multimodal time series foundation model that handles numbers and text together and forecasts zero-shot. Paper by Wang et al. (2024).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2412.11376", "https://huggingface.co/ChengsenWang/ChatTime-1-7B-Chat", "https://ai.azure.com/catalog/models?task=time-series-forecasting" ], "unverified": [] }, { "id": "chattime-1-7b-base", "name": "ChatTime-1-7B-Base", "level": "variant", "family_id": "chattime", "developer": [], "model_class": "time series", "modalities_in": [ "time series", "text" ], "modalities_out": [ "time series", "text" ], "architecture": "autoregressive", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-07-08", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Base ChatTime checkpoint, fine-tuned from Llama 2 7B according to its Hugging Face base_model field.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/ChengsenWang/ChatTime-1-7B-Base" ], "unverified": [] }, { "id": "chattime-1-7b-chat", "name": "ChatTime-1-7B-Chat", "level": "variant", "family_id": "chattime", "developer": [], "model_class": "time series", "modalities_in": [ "time series", "text" ], "modalities_out": [ "time series", "text" ], "architecture": "autoregressive", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-07-08", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Chat ChatTime checkpoint, fine-tuned from ChatTime-1-7B-Base, for forecasting and question answering over time series and text.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/ChengsenWang/ChatTime-1-7B-Chat", "https://ai.azure.com/catalog/models?task=time-series-forecasting" ], "unverified": [] }, { "id": "anomaly-transformer", "name": "Anomaly Transformer", "level": "family", "family_id": null, "developer": [ "tsinghua-university" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "dense transformer", "architecture_detail": "Association discrepancy between prior and series associations", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-10-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the official thuml/Anomaly-Transformer repository (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Transformer for unsupervised time series anomaly detection (ICLR 2022). Paper by Xu et al. (2021).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2110.02642", "https://github.com/thuml/Anomaly-Transformer", "https://pyod.readthedocs.io/en/latest/" ], "unverified": [] }, { "id": "microsoft-spectral-residual", "name": "Spectral Residual (SR and SR-CNN)", "level": "family", "family_id": null, "developer": [ "microsoft" ], "model_class": "time series", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "CNN", "architecture_detail": "Spectral residual saliency, optionally followed by a CNN", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-06-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the microsoft/anomalydetector repository (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Time series anomaly detection method behind a Microsoft monitoring service, from the paper Time-Series Anomaly Detection Service at Microsoft (Ren et al., KDD 2019).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1906.03821", "https://github.com/microsoft/anomalydetector", "https://pyod.readthedocs.io/en/latest/" ], "unverified": [] }, { "id": "gcn", "name": "GCN (Graph Convolutional Network)", "level": "family", "family_id": null, "developer": [ "university-of-amsterdam" ], "model_class": "graph", "modalities_in": [ "graph" ], "modalities_out": [ "graph" ], "architecture": "GNN", "architecture_detail": "Layer-wise first-order spectral graph convolution", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2016-09-09", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the authors' tkipf/gcn repository (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Graph convolutional network for semi-supervised node classification, a basic building block of graph learning. Paper by Kipf and Welling (2016).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1609.02907", "https://github.com/tkipf/gcn", "https://github.com/pyg-team/pytorch_geometric", "https://pytorch-geometric.readthedocs.io/en/latest/modules/nn.html", "https://modelzoo.co/model/pygcn", "https://huggingface.co/datasets/pwc-archive/methods", "https://arxiv.org/html/1609.02907" ], "unverified": [] }, { "id": "gat", "name": "GAT (Graph Attention Network)", "level": "family", "family_id": null, "developer": [ "university-of-cambridge", "mila" ], "model_class": "graph", "modalities_in": [ "graph" ], "modalities_out": [ "graph" ], "architecture": "GNN", "architecture_detail": "Masked self-attention over neighbours", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-10-30", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the authors' PetarV-/GAT repository (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Graph neural network that weights neighbours with attention. Paper by Velickovic et al. (2017).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1710.10903", "https://github.com/PetarV-/GAT", "https://github.com/pyg-team/pytorch_geometric", "https://pytorch-geometric.readthedocs.io/en/latest/modules/nn.html", "https://modelzoo.co/model/pygat", "https://huggingface.co/datasets/pwc-archive/methods", "https://arxiv.org/html/1710.10903" ], "unverified": [] }, { "id": "gatv2", "name": "GATv2", "level": "variant", "family_id": "gat", "developer": [ "technion" ], "model_class": "graph", "modalities_in": [ "graph" ], "modalities_out": [ "graph" ], "architecture": "GNN", "architecture_detail": "Dynamic attention variant of GAT", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-05-30", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the PyTorch Geometric implementation (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Revised graph attention layer that computes dynamic rather than static attention, from the paper How Attentive are Graph Attention Networks? (Brody et al., 2021).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2105.14491", "https://github.com/pyg-team/pytorch_geometric", "https://pytorch-geometric.readthedocs.io/en/latest/modules/nn.html", "https://arxiv.org/html/2105.14491" ], "unverified": [] }, { "id": "graphsage", "name": "GraphSAGE", "level": "family", "family_id": null, "developer": [ "stanford-university" ], "model_class": "graph", "modalities_in": [ "graph" ], "modalities_out": [ "graph" ], "architecture": "GNN", "architecture_detail": "Inductive neighbourhood sampling and aggregation", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-06-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": "Reference repository williamleif/GraphSAGE has no standard licence identified by GitHub; PyTorch Geometric implementation is MIT. ONNX Model Zoo publishes SAGEConv graph models.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Inductive method that learns node embeddings by sampling and aggregating features from neighbours, so it works on unseen nodes and large graphs. Paper by Hamilton et al. (2017).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1706.02216", "https://github.com/williamleif/GraphSAGE", "https://github.com/pyg-team/pytorch_geometric", "https://pytorch-geometric.readthedocs.io/en/latest/modules/nn.html", "https://github.com/onnx/models/tree/main/Graph_Machine_Learning/sageconv_Opset16_graph_convolutions", "https://huggingface.co/datasets/pwc-archive/methods", "https://arxiv.org/html/1706.02216" ], "unverified": [] }, { "id": "gin", "name": "GIN (Graph Isomorphism Network)", "level": "family", "family_id": null, "developer": [ "massachusetts-institute-of-technology", "stanford-university" ], "model_class": "graph", "modalities_in": [ "graph" ], "modalities_out": [ "graph" ], "architecture": "GNN", "architecture_detail": "Sum aggregation with MLPs, as expressive as the Weisfeiler-Lehman test", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-10-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the PyTorch Geometric implementation (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Graph neural network designed for maximal expressive power among message-passing networks, from the paper How Powerful are Graph Neural Networks? (Xu et al., 2018).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1810.00826", "https://github.com/pyg-team/pytorch_geometric", "https://pytorch-geometric.readthedocs.io/en/latest/modules/nn.html", "https://huggingface.co/datasets/pwc-archive/methods", "https://arxiv.org/html/1810.00826" ], "unverified": [] }, { "id": "node2vec", "name": "node2vec", "level": "family", "family_id": null, "developer": [ "stanford-university" ], "model_class": "graph", "modalities_in": [ "graph" ], "modalities_out": [ "graph" ], "architecture": "other", "architecture_detail": "Biased random walks plus skip-gram node embeddings", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2016-07-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the authors' aditya-grover/node2vec repository (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Scalable method for learning node embeddings from biased random walks over a network. Paper by Grover and Leskovec (2016).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1607.00653", "https://github.com/aditya-grover/node2vec", "https://github.com/pyg-team/pytorch_geometric", "https://pytorch-geometric.readthedocs.io/en/latest/modules/nn.html", "https://huggingface.co/datasets/pwc-archive/methods", "https://arxiv.org/html/1607.00653" ], "unverified": [] }, { "id": "deepwalk", "name": "DeepWalk", "level": "family", "family_id": null, "developer": [ "stony-brook-university" ], "model_class": "graph", "modalities_in": [ "graph" ], "modalities_out": [ "graph" ], "architecture": "other", "architecture_detail": "Truncated random walks plus skip-gram node embeddings", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2014-03-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": "Reference repository phanein/deepwalk has no standard licence identified by GitHub.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Early method for learning node representations from random walks on a graph. Paper by Perozzi et al. (2014).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1403.6652", "https://github.com/phanein/deepwalk", "https://www.dgl.ai/dgl_docs/api/python/nn-pytorch.html", "https://huggingface.co/datasets/pwc-archive/methods", "https://arxiv.org/html/1403.6652" ], "unverified": [] }, { "id": "tgn", "name": "TGN (Temporal Graph Networks)", "level": "family", "family_id": null, "developer": [ "twitter" ], "model_class": "graph", "modalities_in": [ "graph" ], "modalities_out": [ "graph" ], "architecture": "GNN", "architecture_detail": "Memory modules plus graph operators for continuous-time dynamic graphs", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-06-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the twitter-research/tgn repository (Apache-2.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Framework for deep learning on dynamic graphs whose edges arrive as timed events. Paper by Rossi et al. (2020).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2006.10637", "https://github.com/twitter-research/tgn", "https://github.com/pyg-team/pytorch_geometric", "https://pytorch-geometric.readthedocs.io/en/latest/modules/nn.html", "https://huggingface.co/datasets/pwc-archive/methods", "https://arxiv.org/html/2006.10637" ], "unverified": [] }, { "id": "r-gcn", "name": "R-GCN (Relational Graph Convolutional Network)", "level": "family", "family_id": null, "developer": [ "university-of-amsterdam" ], "model_class": "graph", "modalities_in": [ "graph" ], "modalities_out": [ "graph" ], "architecture": "GNN", "architecture_detail": "Relation-specific graph convolution", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-03-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the PyTorch Geometric implementation (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Graph convolutional network for relational data such as knowledge graphs, for link prediction and entity classification. Paper by Schlichtkrull et al. (2017).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1703.06103", "https://github.com/pyg-team/pytorch_geometric", "https://pytorch-geometric.readthedocs.io/en/latest/modules/nn.html", "https://arxiv.org/html/1703.06103" ], "unverified": [] }, { "id": "hgt", "name": "HGT (Heterogeneous Graph Transformer)", "level": "family", "family_id": null, "developer": [ "ucla", "microsoft-research" ], "model_class": "graph", "modalities_in": [ "graph" ], "modalities_out": [ "graph" ], "architecture": "GNN", "architecture_detail": "Type-dependent attention for heterogeneous graphs", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-03-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the PyTorch Geometric implementation (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Transformer-style graph neural network for graphs with many node and edge types. Paper by Hu et al. (2020).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2003.01332", "https://github.com/pyg-team/pytorch_geometric", "https://pytorch-geometric.readthedocs.io/en/latest/modules/nn.html", "https://arxiv.org/html/2003.01332" ], "unverified": [] }, { "id": "graphgps", "name": "GraphGPS", "level": "family", "family_id": null, "developer": [ "mila", "universite-de-montreal" ], "model_class": "graph", "modalities_in": [ "graph" ], "modalities_out": [ "graph" ], "architecture": "GNN", "architecture_detail": "Local message passing combined with global attention", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-05-25", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the authors' rampasek/GraphGPS repository (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Recipe for general, powerful and scalable graph transformers that combine message passing with global attention. Paper by Rampasek et al. (2022).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2205.12454", "https://github.com/rampasek/GraphGPS", "https://github.com/pyg-team/pytorch_geometric", "https://pytorch-geometric.readthedocs.io/en/latest/modules/nn.html", "https://arxiv.org/html/2205.12454" ], "unverified": [] }, { "id": "microsoft-graphormer", "name": "Graphormer", "level": "family", "family_id": null, "developer": [ "microsoft" ], "model_class": "graph", "modalities_in": [ "graph" ], "modalities_out": [ "graph" ], "architecture": "dense transformer", "architecture_detail": "Transformer with structural encodings of graphs", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-06-09", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the microsoft/Graphormer repository (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Graph transformer from Microsoft, positioned by its repository as a general deep learning backbone for molecular modelling. Paper by Ying et al. (2021).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2106.05234", "https://github.com/microsoft/Graphormer", "https://www.dgl.ai/dgl_docs/api/python/nn-pytorch.html" ], "unverified": [] }, { "id": "microsoft-graphormer-base-pcqm4mv2", "name": "Graphormer base (PCQM4Mv2)", "level": "variant", "family_id": "microsoft-graphormer", "developer": [ "microsoft" ], "model_class": "graph", "modalities_in": [ "graph" ], "modalities_out": [ "graph" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-01-05", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": "Hosted on Hugging Face by the user clefourrier; the model card names Microsoft as developer.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Graphormer base checkpoint pre-trained on PCQM4M-LSCv2 for graph classification.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/clefourrier/graphormer-base-pcqm4mv2", "https://huggingface.co/api/models?pipeline_tag=graph-ml&sort=downloads&limit=50" ], "unverified": [] }, { "id": "ultra-kg", "name": "ULTRA", "level": "family", "family_id": null, "developer": [ "intel", "mila" ], "model_class": "graph", "modalities_in": [ "graph" ], "modalities_out": [ "graph" ], "architecture": "GNN", "architecture_detail": "Relative relation representations so one model transfers to any knowledge graph", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints pre-trained on 3, 4 and 50 graphs; ultra_50g is the 50-graph one (model card).", "context_length": null, "release_date": "2023-10-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Foundation model for knowledge graph reasoning: a single pre-trained model predicts links on any multi-relational graph, zero-shot or after fine-tuning. Paper by Galkin et al. (2023).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2310.04562", "https://github.com/DeepGraphLearning/ULTRA", "https://huggingface.co/mgalkin/ultra_50g", "https://arxiv.org/html/2310.04562" ], "unverified": [ "params_note" ] }, { "id": "microsoft-gridsfm-open", "name": "GridSFM Open", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "graph", "modalities_in": [ "graph" ], "modalities_out": [ "graph" ], "architecture": "GNN", "architecture_detail": "Heterogeneous graph neural network over power grid representations (model card)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-05-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft Grid Small Foundation Model: a graph neural network trained on simulated power grid scenarios to learn how topology and load relate to system behaviour, so it generalises across grid configurations.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/microsoft/GridSFM_Open", "https://huggingface.co/api/models?pipeline_tag=graph-ml&sort=likes&limit=50" ], "unverified": [] }, { "id": "rotate-kge", "name": "RotatE", "level": "family", "family_id": null, "developer": [ "peking-university", "mila" ], "model_class": "graph", "modalities_in": [ "graph" ], "modalities_out": [ "graph" ], "architecture": "other", "architecture_detail": "Relations as rotations in complex vector space", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-02-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the PyTorch Geometric implementation (MIT). TransE, the earlier translation-based embedding, is listed with it by Papers with Code.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Knowledge graph embedding that models each relation as a rotation in complex space, for link prediction (ICLR 2019). Paper by Sun et al. (2019).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1902.10197", "https://github.com/pyg-team/pytorch_geometric", "https://pytorch-geometric.readthedocs.io/en/latest/modules/nn.html", "https://huggingface.co/datasets/pwc-archive/methods", "https://arxiv.org/html/1902.10197" ], "unverified": [] }, { "id": "schnet", "name": "SchNet", "level": "family", "family_id": null, "developer": [ "tu-berlin" ], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "GNN", "architecture_detail": "Continuous-filter convolutions over atom positions", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-06-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the PyTorch Geometric implementation (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Continuous-filter convolutional network for modelling quantum interactions in molecules, such as energies and forces. Paper by Schutt et al. (2017).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1706.08566", "https://github.com/pyg-team/pytorch_geometric", "https://pytorch-geometric.readthedocs.io/en/latest/modules/nn.html", "https://huggingface.co/datasets/pwc-archive/methods", "https://arxiv.org/html/1706.08566" ], "unverified": [] }, { "id": "dimenet", "name": "DimeNet and DimeNet++", "level": "family", "family_id": null, "developer": [ "tum" ], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "GNN", "architecture_detail": "Directional message passing using bond angles", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-03-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the PyTorch Geometric implementation (MIT). DimeNet++ (arXiv 2011.14115, 28 Nov 2020) is the faster successor.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Directional message passing networks for molecular graphs that use distances and angles between atoms. Paper by Gasteiger et al. (2020).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2003.03123", "https://arxiv.org/abs/2011.14115", "https://github.com/pyg-team/pytorch_geometric", "https://pytorch-geometric.readthedocs.io/en/latest/modules/nn.html", "https://arxiv.org/html/2003.03123" ], "unverified": [] }, { "id": "visnet", "name": "ViSNet", "level": "family", "family_id": null, "developer": [ "microsoft-research" ], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "GNN", "architecture_detail": "Equivariant vector-scalar interactive message passing", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-10-29", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the PyTorch Geometric implementation (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Equivariant geometry-enhanced graph neural network for molecular property and dynamics prediction. Paper by Wang et al. (2022).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2210.16518", "https://github.com/pyg-team/pytorch_geometric", "https://pytorch-geometric.readthedocs.io/en/latest/modules/nn.html", "https://arxiv.org/html/2210.16518" ], "unverified": [] }, { "id": "egnn", "name": "EGNN (E(n) Equivariant Graph Neural Network)", "level": "family", "family_id": null, "developer": [ "university-of-amsterdam" ], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "GNN", "architecture_detail": "Equivariant to rotations, translations, reflections and permutations", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-02-19", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the authors' vgsatorras/egnn repository (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Graph neural network equivariant to Euclidean transformations, used for molecules and physical systems. Paper by Garcia Satorras et al. (2021).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2102.09844", "https://github.com/vgsatorras/egnn", "https://www.dgl.ai/dgl_docs/api/python/nn-pytorch.html", "https://arxiv.org/html/2102.09844" ], "unverified": [] }, { "id": "se3-transformer", "name": "SE(3)-Transformer", "level": "family", "family_id": null, "developer": [ "university-of-oxford", "bosch" ], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "GNN", "architecture_detail": "Roto-translation equivariant attention on 3D point clouds and graphs", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-06-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Self-attention module for 3D point clouds and graphs that is equivariant to rotations and translations. Paper by Fuchs et al. (2020); NVIDIA Deep Learning Examples publish trained checkpoints on NGC.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2006.10503", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/dle/models/se3transformer__pyt_ckpt_mode-homo", "https://arxiv.org/html/2006.10503" ], "unverified": [] }, { "id": "google-wide-and-deep", "name": "Wide and Deep", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "recommendation", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": "Jointly trained wide linear model and deep neural network", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2016-06-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": "NVIDIA Deep Learning Examples publish TensorFlow checkpoints trained on Outbrain data on NGC.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Recommender architecture that trains a wide linear model and a deep network together, productionised on Google Play. Paper by Cheng et al. (2016).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1606.07792", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/dle/models/widendeep_base_tf1_ckpt", "https://github.com/RUCAIBox/RecBole2.0/blob/main/model_list.md" ], "unverified": [] }, { "id": "deepfm", "name": "DeepFM", "level": "family", "family_id": null, "developer": [ "harbin-institute-of-technology", "huawei-technologies" ], "model_class": "recommendation", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": "Factorisation machine and deep network sharing embeddings", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-03-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the RecBole implementation (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Factorisation-machine based neural network for click-through rate prediction. Paper by Guo et al. (2017).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1703.04247", "https://github.com/RUCAIBox/RecBole", "https://github.com/RUCAIBox/RecBole2.0/blob/main/model_list.md", "https://arxiv.org/html/1703.04247" ], "unverified": [] }, { "id": "xdeepfm", "name": "xDeepFM", "level": "family", "family_id": null, "developer": [ "university-of-science-and-technology-of-china", "microsoft-research" ], "model_class": "recommendation", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": "Compressed interaction network plus deep network", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-03-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the RecBole implementation (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Recommender model that combines explicit and implicit feature interactions. Paper by Lian et al. (2018).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1803.05170", "https://github.com/RUCAIBox/RecBole", "https://github.com/RUCAIBox/RecBole2.0/blob/main/model_list.md", "https://arxiv.org/html/1803.05170" ], "unverified": [] }, { "id": "alibaba-din", "name": "DIN (Deep Interest Network)", "level": "family", "family_id": null, "developer": [ "alibaba" ], "model_class": "recommendation", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": "Attention over user behaviour history relative to the candidate ad", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-06-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the RecBole implementation (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Click-through rate model that attends to the parts of a user's history relevant to each candidate, deployed in Alibaba's online display advertising. Paper by Zhou et al. (2017).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1706.06978", "https://github.com/RUCAIBox/RecBole", "https://github.com/RUCAIBox/RecBole2.0/blob/main/model_list.md" ], "unverified": [] }, { "id": "dien", "name": "DIEN (Deep Interest Evolution Network)", "level": "family", "family_id": null, "developer": [ "alibaba-group" ], "model_class": "recommendation", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": "Models how user interests evolve over time", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-09-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the RecBole implementation (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Click-through rate model that captures the evolution of user interests, deployed in Taobao display advertising. Paper by Zhou et al. (2018).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1809.03672", "https://github.com/RUCAIBox/RecBole", "https://github.com/RUCAIBox/RecBole2.0/blob/main/model_list.md", "https://arxiv.org/html/1809.03672" ], "unverified": [] }, { "id": "alibaba-sim", "name": "SIM (Search-based Interest Model)", "level": "family", "family_id": null, "developer": [ "alibaba" ], "model_class": "recommendation", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": "Search-based retrieval over lifelong user behaviour sequences", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-06-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": "NVIDIA Deep Learning Examples publish a TensorFlow 2 checkpoint trained on Amazon Books on NGC.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Click-through rate model for very long user behaviour histories, deployed in Alibaba display advertising since 2019. Paper by Qi et al. (2020).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2006.05639", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/dle/models/sim__tf2_ckpt_ds-amazonbooks" ], "unverified": [] }, { "id": "dcn", "name": "DCN (Deep and Cross Network)", "level": "family", "family_id": null, "developer": [ "google", "stanford-university" ], "model_class": "recommendation", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": "Explicit feature crossing layers plus deep network", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-08-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the RecBole implementation (MIT). AURION also holds dlrm-dcnv2, the MLPerf training workload.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Network that learns bounded-degree feature crosses for ad click prediction. Paper by Wang et al. (2017).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1708.05123", "https://github.com/RUCAIBox/RecBole", "https://github.com/RUCAIBox/RecBole2.0/blob/main/model_list.md", "https://arxiv.org/html/1708.05123" ], "unverified": [] }, { "id": "google-dcn-v2", "name": "DCN V2", "level": "variant", "family_id": "dcn", "developer": [ "google" ], "model_class": "recommendation", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-08-19", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Improved Deep and Cross Network, used across many web-scale learning-to-rank systems at Google. Paper by Wang et al. (2020).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2008.13535", "https://nvidia-merlin.github.io/models/stable/models_overview.html", "https://github.com/RUCAIBox/RecBole2.0/blob/main/model_list.md" ], "unverified": [] }, { "id": "meta-dlrm", "name": "DLRM (Deep Learning Recommendation Model)", "level": "family", "family_id": null, "developer": [ "meta" ], "model_class": "recommendation", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": "Embedding tables for sparse features plus MLPs and pairwise feature interactions", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-05-31", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the facebookresearch/dlrm repository (MIT). AURION already holds dlrm-dcnv2, the MLPerf variant.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Neural recommendation model for personalisation that handles categorical features through embedding tables; its implementation is published by Facebook Research (facebookresearch/dlrm). Paper by Naumov et al. (2019).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1906.00091", "https://github.com/facebookresearch/dlrm", "https://nvidia-merlin.github.io/models/stable/models_overview.html", "https://github.com/meta-pytorch/torchrec" ], "unverified": [] }, { "id": "sasrec", "name": "SASRec", "level": "family", "family_id": null, "developer": [ "uc-san-diego" ], "model_class": "recommendation", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "dense transformer", "architecture_detail": "Causal self-attention over the item sequence", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-08-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the authors' kang205/SASRec repository (Apache-2.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Self-attentive sequential recommendation model that predicts the next item from a user's item history. Paper by Kang and McAuley (2018).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1808.09781", "https://github.com/kang205/SASRec", "https://github.com/RUCAIBox/RecBole2.0/blob/main/model_list.md", "https://arxiv.org/html/1808.09781" ], "unverified": [] }, { "id": "bert4rec", "name": "BERT4Rec", "level": "family", "family_id": null, "developer": [ "alibaba-group" ], "model_class": "recommendation", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "dense transformer", "architecture_detail": "Bidirectional transformer trained with masked item prediction", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-04-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the authors' FeiSun/BERT4Rec repository (Apache-2.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Sequential recommendation model using bidirectional transformer encoders. Paper by Sun et al. (2019).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1904.06690", "https://github.com/FeiSun/BERT4Rec", "https://github.com/RUCAIBox/RecBole2.0/blob/main/model_list.md", "https://arxiv.org/html/1904.06690" ], "unverified": [] }, { "id": "gru4rec", "name": "GRU4Rec", "level": "family", "family_id": null, "developer": [ "gravity-rd" ], "model_class": "recommendation", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": "Recurrent network with GRU cells over session clicks", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2015-11-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": "Reference repository hidasib/GRU4Rec has no standard licence identified by GitHub; RecBole implementation is MIT.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Session-based recommendation with recurrent neural networks. Paper by Hidasi et al. (2015).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1511.06939", "https://github.com/hidasib/GRU4Rec", "https://github.com/RUCAIBox/RecBole2.0/blob/main/model_list.md", "https://arxiv.org/html/1511.06939" ], "unverified": [] }, { "id": "ncf", "name": "NCF (Neural Collaborative Filtering) and NeuMF", "level": "family", "family_id": null, "developer": [ "nus" ], "model_class": "recommendation", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": "Neural matrix factorisation combining generalised matrix factorisation and an MLP", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-08-16", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the RecBole implementation (MIT). Also in TensorFlow Model Garden and ONNX Model Zoo.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Neural network framework for collaborative filtering on implicit feedback. Paper by He et al. (2017).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1708.05031", "https://github.com/RUCAIBox/RecBole", "https://github.com/RUCAIBox/RecBole2.0/blob/main/model_list.md", "https://github.com/tensorflow/models/blob/master/official/README.md", "https://github.com/onnx/models/blob/main/README.md", "https://arxiv.org/html/1708.05031" ], "unverified": [] }, { "id": "ngcf", "name": "NGCF (Neural Graph Collaborative Filtering)", "level": "family", "family_id": null, "developer": [ "nus", "university-of-science-and-technology-of-china" ], "model_class": "recommendation", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "GNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-05-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the RecBole implementation (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Collaborative filtering that propagates embeddings over the user-item graph. Paper by Wang et al. (2019).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1905.08108", "https://github.com/RUCAIBox/RecBole", "https://github.com/RUCAIBox/RecBole2.0/blob/main/model_list.md", "https://arxiv.org/html/1905.08108" ], "unverified": [] }, { "id": "lightgcn", "name": "LightGCN", "level": "family", "family_id": null, "developer": [ "university-of-science-and-technology-of-china" ], "model_class": "recommendation", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "GNN", "architecture_detail": "Graph convolution reduced to neighbourhood aggregation", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-02-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the RecBole implementation (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Simplified graph convolution network for collaborative filtering recommendation. Paper by He et al. (2020).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2002.02126", "https://github.com/RUCAIBox/RecBole", "https://github.com/RUCAIBox/RecBole2.0/blob/main/model_list.md", "https://huggingface.co/datasets/pwc-archive/methods", "https://arxiv.org/html/2002.02126" ], "unverified": [] }, { "id": "multvae", "name": "Mult-VAE (variational autoencoder for collaborative filtering)", "level": "family", "family_id": null, "developer": [ "netflix" ], "model_class": "recommendation", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-02-16", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the RecBole implementation (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Variational autoencoder with a multinomial likelihood for collaborative filtering on implicit feedback. Paper by Liang et al. (2018).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1802.05814", "https://github.com/RUCAIBox/RecBole", "https://github.com/RUCAIBox/RecBole2.0/blob/main/model_list.md", "https://arxiv.org/html/1802.05814" ], "unverified": [] }, { "id": "bpr", "name": "BPR (Bayesian Personalized Ranking)", "level": "family", "family_id": null, "developer": [], "model_class": "recommendation", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "other", "architecture_detail": "Pairwise ranking objective for implicit feedback", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2012-05-09", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "arXiv date is the 2012 posting of the UAI 2009 paper. Licence is that of the RecBole implementation (MIT).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Pairwise ranking method for recommendation from implicit feedback, a standard baseline. Paper by Rendle et al.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1205.2618", "https://github.com/RUCAIBox/RecBole", "https://github.com/RUCAIBox/RecBole2.0/blob/main/model_list.md" ], "unverified": [] }, { "id": "mmoe", "name": "MMoE (Multi-gate Mixture-of-Experts)", "level": "family", "family_id": null, "developer": [], "model_class": "recommendation", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "MoE", "architecture_detail": "Shared experts with one gating network per task", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multi-task learning architecture with a gate per task over shared experts, used in recommender systems. Paper by Ma et al. (KDD 2018).", "verified_on": "2026-10-12", "sources": [ "https://doi.org/10.1145/3219819.3220007", "https://nvidia-merlin.github.io/models/stable/models_overview.html" ], "unverified": [] }, { "id": "ple", "name": "PLE (Progressive Layered Extraction)", "level": "family", "family_id": null, "developer": [], "model_class": "recommendation", "modalities_in": [ "table" ], "modalities_out": [ "table" ], "architecture": "MoE", "architecture_detail": "Separates shared and task-specific experts across layers", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020", "release_date_basis": "notes", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multi-task learning model for recommendation that separates shared and task-specific experts. Paper by Tang et al. (RecSys 2020).", "verified_on": "2026-10-12", "sources": [ "https://doi.org/10.1145/3383313.3412236", "https://nvidia-merlin.github.io/models/stable/models_overview.html" ], "unverified": [] }, { "id": "anomalib", "name": "Anomalib", "level": "family", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "other", "architecture_detail": "Library of anomaly detection algorithms with training, benchmarking and deployment tools", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-02-16", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open library of visual anomaly detection algorithms with experiment management and deployment tools. Paper by Akcay et al. (2022).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2202.08341", "https://github.com/open-edge-platform/anomalib", "https://anomalib.readthedocs.io/en/latest/markdown/guides/reference/models/image/index.html" ], "unverified": [] }, { "id": "amazon-patchcore", "name": "PatchCore", "level": "family", "family_id": null, "developer": [ "amazon" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "other", "architecture_detail": "Memory bank of nominal patch features with coreset subsampling", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-06-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the amazon-science/patchcore-inspection repository (Apache-2.0). Also in Anomalib.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Industrial visual anomaly detector that compares image patches with a memory bank of normal patch features, from the paper Towards Total Recall in Industrial Anomaly Detection (Roth et al., CVPR 2022).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2106.08265", "https://github.com/amazon-science/patchcore-inspection", "https://github.com/open-edge-platform/anomalib", "https://anomalib.readthedocs.io/en/latest/markdown/guides/reference/models/image/index.html" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "padim", "name": "PaDiM", "level": "family", "family_id": null, "developer": [ "cea" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "other", "architecture_detail": "Gaussian model of pre-trained CNN patch embeddings", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-11-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the Anomalib implementation (Apache-2.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Patch distribution modelling framework for detecting and localising anomalies in images. Paper by Defard et al. (2020).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2011.08785", "https://github.com/open-edge-platform/anomalib", "https://anomalib.readthedocs.io/en/latest/markdown/guides/reference/models/image/index.html", "https://arxiv.org/html/2011.08785" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "efficientad", "name": "EfficientAD", "level": "family", "family_id": null, "developer": [ "mvtec" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": "Lightweight student-teacher plus autoencoder", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-03-25", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the Anomalib implementation (Apache-2.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Visual anomaly detector built for millisecond-level latency. Paper by Batzner et al. (2023).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2303.14535", "https://github.com/open-edge-platform/anomalib", "https://anomalib.readthedocs.io/en/latest/markdown/guides/reference/models/image/index.html", "https://arxiv.org/html/2303.14535" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "fastflow", "name": "FastFlow", "level": "family", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "other", "architecture_detail": "2D normalising flows on CNN features", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-11-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the Anomalib implementation (Apache-2.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Unsupervised anomaly detection and localisation with 2D normalising flows. Paper by Yu et al. (2021).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2111.07677", "https://github.com/open-edge-platform/anomalib", "https://anomalib.readthedocs.io/en/latest/markdown/guides/reference/models/image/index.html" ], "unverified": [] }, { "id": "draem", "name": "DRAEM", "level": "family", "family_id": null, "developer": [ "university-of-ljubljana" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": "Reconstructive and discriminative subnetworks trained on simulated anomalies", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-08-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the Anomalib implementation (Apache-2.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Discriminatively trained reconstruction model for surface anomaly detection. Paper by Zavrtanik et al. (2021).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2108.07610", "https://github.com/open-edge-platform/anomalib", "https://anomalib.readthedocs.io/en/latest/markdown/guides/reference/models/image/index.html", "https://arxiv.org/html/2108.07610" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "stfpm", "name": "STFPM", "level": "family", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": "Student-teacher feature pyramid matching", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-03-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the Anomalib implementation (Apache-2.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Student-teacher feature pyramid matching for anomaly detection. Paper by Wang et al. (2021).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.04257", "https://github.com/open-edge-platform/anomalib", "https://anomalib.readthedocs.io/en/latest/markdown/guides/reference/models/image/index.html" ], "unverified": [] }, { "id": "dinomaly", "name": "Dinomaly", "level": "family", "family_id": null, "developer": [ "beijing-institute-of-technology" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": "Reconstruction on DINO vision transformer features", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-05-23", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the Anomalib implementation (Apache-2.0). AURION holds its successor Dinomaly2 (tsinghua-university-dinomaly2) separately.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multi-class unsupervised anomaly detection with a deliberately simple transformer reconstruction design. Paper by Guo et al. (2024).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2405.14325", "https://github.com/open-edge-platform/anomalib", "https://anomalib.readthedocs.io/en/latest/markdown/guides/reference/models/image/index.html", "https://arxiv.org/html/2405.14325" ], "unverified": [] }, { "id": "winclip", "name": "WinCLIP", "level": "family", "family_id": null, "developer": [ "amazon", "kaist" ], "model_class": "vision", "modalities_in": [ "image", "text" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": "Window-based CLIP features compared with text prompts", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-03-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the Anomalib implementation (Apache-2.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Zero-shot and few-shot anomaly classification and segmentation using CLIP. Paper by Jeong et al. (2023).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2303.14814", "https://github.com/open-edge-platform/anomalib", "https://anomalib.readthedocs.io/en/latest/markdown/guides/reference/models/image/index.html", "https://arxiv.org/html/2303.14814" ], "unverified": [] }, { "id": "reverse-distillation", "name": "Reverse Distillation", "level": "family", "family_id": null, "developer": [ "university-of-alberta" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": "Student decoder reconstructs teacher features from a one-class embedding", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-01-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the Anomalib implementation (Apache-2.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Anomaly detection by reverse knowledge distillation from a one-class embedding. Paper by Deng and Li (2022).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2201.10703", "https://github.com/open-edge-platform/anomalib", "https://anomalib.readthedocs.io/en/latest/markdown/guides/reference/models/image/index.html", "https://arxiv.org/html/2201.10703" ], "unverified": [] }, { "id": "cfa", "name": "CFA", "level": "family", "family_id": null, "developer": [ "inha-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "other", "architecture_detail": "Coupled-hypersphere feature adaptation", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-06-09", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the Anomalib implementation (Apache-2.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Coupled-hypersphere-based feature adaptation for target-oriented anomaly localisation. Paper by Lee et al. (2022).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2206.04325", "https://github.com/open-edge-platform/anomalib", "https://anomalib.readthedocs.io/en/latest/markdown/guides/reference/models/image/index.html", "https://arxiv.org/html/2206.04325" ], "unverified": [] }, { "id": "cs-flow", "name": "CS-Flow", "level": "family", "family_id": null, "developer": [ "leibniz-university-hannover" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "other", "architecture_detail": "Fully convolutional cross-scale normalising flows", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-10-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the Anomalib implementation (Apache-2.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Cross-scale normalising flows for image-based defect detection. Paper by Rudolph et al. (2021).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2110.02855", "https://github.com/open-edge-platform/anomalib", "https://anomalib.readthedocs.io/en/latest/markdown/guides/reference/models/image/index.html", "https://arxiv.org/html/2110.02855" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "ganomaly", "name": "GANomaly", "level": "family", "family_id": null, "developer": [ "durham-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "other", "architecture_detail": "Encoder-decoder-encoder GAN", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-05-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the Anomalib implementation (Apache-2.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Semi-supervised anomaly detection via adversarial training. Paper by Akcay et al. (2018).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1805.06725", "https://github.com/open-edge-platform/anomalib", "https://anomalib.readthedocs.io/en/latest/markdown/guides/reference/models/image/index.html", "https://arxiv.org/html/1805.06725" ], "unverified": [] }, { "id": "anomalydino", "name": "AnomalyDINO", "level": "family", "family_id": null, "developer": [ "ruhr-university-bochum" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": "Patch nearest-neighbour matching on DINOv2 features", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-05-23", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the Anomalib implementation (Apache-2.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Few-shot patch-based anomaly detection using DINOv2 features. Paper by Damm et al. (2024).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2405.14529", "https://github.com/open-edge-platform/anomalib", "https://anomalib.readthedocs.io/en/latest/markdown/guides/reference/models/image/index.html", "https://arxiv.org/html/2405.14529" ], "unverified": [] }, { "id": "supersimplenet", "name": "SuperSimpleNet", "level": "family", "family_id": null, "developer": [ "university-of-ljubljana" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-08-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the Anomalib implementation (Apache-2.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fast surface defect detector that unifies unsupervised and supervised learning. Paper by Rolih et al. (2024).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2408.03143", "https://github.com/open-edge-platform/anomalib", "https://anomalib.readthedocs.io/en/latest/markdown/guides/reference/models/image/index.html", "https://arxiv.org/html/2408.03143" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "glass-anomaly", "name": "GLASS", "level": "family", "family_id": null, "developer": [ "casia" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "other", "architecture_detail": "Anomaly synthesis with gradient ascent", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-07-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the Anomalib implementation (Apache-2.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Unified anomaly synthesis strategy for industrial anomaly detection and localisation. Paper by Chen et al. (2024).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2407.09359", "https://github.com/open-edge-platform/anomalib", "https://anomalib.readthedocs.io/en/latest/markdown/guides/reference/models/image/index.html", "https://arxiv.org/html/2407.09359" ], "unverified": [ "indonesian_measured", "indonesia_hosting" ] }, { "id": "ppo", "name": "PPO (Proximal Policy Optimization)", "level": "family", "family_id": null, "developer": [], "model_class": "reinforcement learning", "modalities_in": [], "modalities_out": [ "action" ], "architecture": "other", "architecture_detail": "Policy gradient with a clipped surrogate objective", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-07-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the Stable-Baselines3 implementation (MIT). model_class left empty: AURION has no reinforcement learning class yet.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "On-policy reinforcement learning algorithm that improves a policy in small steps by clipping its surrogate objective. Paper by Schulman et al. (2017).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1707.06347", "https://github.com/DLR-RM/stable-baselines3", "https://stable-baselines3.readthedocs.io/en/master/guide/algos.html", "https://docs.ray.io/en/latest/rllib/rllib-algorithms.html" ], "unverified": [] }, { "id": "sac", "name": "SAC (Soft Actor-Critic)", "level": "family", "family_id": null, "developer": [ "uc-berkeley" ], "model_class": "reinforcement learning", "modalities_in": [], "modalities_out": [ "action" ], "architecture": "other", "architecture_detail": "Off-policy maximum entropy actor-critic", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-01-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the Stable-Baselines3 implementation (MIT). model_class left empty: AURION has no reinforcement learning class yet.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Off-policy reinforcement learning algorithm that maximises both reward and policy entropy. Paper by Haarnoja et al. (2018).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1801.01290", "https://github.com/DLR-RM/stable-baselines3", "https://stable-baselines3.readthedocs.io/en/master/guide/algos.html", "https://docs.ray.io/en/latest/rllib/rllib-algorithms.html", "https://arxiv.org/html/1801.01290" ], "unverified": [] }, { "id": "dqn", "name": "DQN (Deep Q-Network)", "level": "family", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "reinforcement learning", "modalities_in": [], "modalities_out": [ "action" ], "architecture": "other", "architecture_detail": "Q-learning with a deep network, experience replay and target network", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2013-12-19", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the Stable-Baselines3 implementation (MIT). model_class left empty: AURION has no reinforcement learning class yet.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Deep reinforcement learning method that learns control policies directly from raw pixels, applied to Atari games, from the paper Playing Atari with Deep Reinforcement Learning (Mnih et al., 2013).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1312.5602", "https://github.com/DLR-RM/stable-baselines3", "https://stable-baselines3.readthedocs.io/en/master/guide/algos.html", "https://docs.ray.io/en/latest/rllib/rllib-algorithms.html", "https://arxiv.org/html/1312.5602" ], "unverified": [] }, { "id": "td3", "name": "TD3 (Twin Delayed DDPG)", "level": "family", "family_id": null, "developer": [ "mcgill-university" ], "model_class": "reinforcement learning", "modalities_in": [], "modalities_out": [ "action" ], "architecture": "other", "architecture_detail": "Twin critics, delayed policy updates and target smoothing", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-02-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of the Stable-Baselines3 implementation (MIT). model_class left empty: AURION has no reinforcement learning class yet. DDPG (arXiv 1509.02971), its predecessor, is also in Stable-Baselines3.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Actor-critic algorithm for continuous control that reduces overestimation of value estimates. Paper by Fujimoto et al. (2018).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1802.09477", "https://github.com/DLR-RM/stable-baselines3", "https://stable-baselines3.readthedocs.io/en/master/guide/algos.html", "https://arxiv.org/html/1802.09477" ], "unverified": [] }, { "id": "impala", "name": "IMPALA", "level": "family", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "reinforcement learning", "modalities_in": [], "modalities_out": [ "action" ], "architecture": "other", "architecture_detail": "Distributed actor-learner architecture with V-trace off-policy correction", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-02-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": "Implemented in RLlib (Ray). model_class left empty: AURION has no reinforcement learning class yet.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Scalable distributed deep reinforcement learning architecture with importance-weighted actor-learners. Paper by Espeholt et al. (2018).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1802.01561", "https://docs.ray.io/en/latest/rllib/rllib-algorithms.html", "https://arxiv.org/html/1802.01561" ], "unverified": [] }, { "id": "cql", "name": "CQL (Conservative Q-Learning)", "level": "family", "family_id": null, "developer": [ "uc-berkeley", "google" ], "model_class": "reinforcement learning", "modalities_in": [], "modalities_out": [ "action" ], "architecture": "other", "architecture_detail": "Offline RL with a conservative Q-function lower bound", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-06-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": "Implemented in RLlib (Ray). model_class left empty: AURION has no reinforcement learning class yet.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Offline reinforcement learning method that learns from fixed datasets without further interaction. Paper by Kumar et al. (2020).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2006.04779", "https://docs.ray.io/en/latest/rllib/rllib-algorithms.html", "https://arxiv.org/html/2006.04779" ], "unverified": [] }, { "id": "google-deepmind-alphago", "name": "AlphaGo", "level": "model", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "reinforcement learning", "modalities_in": [], "modalities_out": [ "action" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2015-10", "release_date_basis": "inferred from page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": "model_class left empty: AURION has no reinforcement learning class yet. Date is the first match against Fan Hui (October 2015) per the DeepMind page.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "DeepMind Go-playing system that won the first match between an AI system and a Go professional (Fan Hui, October 2015) and later played Lee Sedol.", "verified_on": "2026-10-12", "sources": [ "https://deepmind.google/research/breakthroughs/alphago/" ], "unverified": [] }, { "id": "google-deepmind-alphazero", "name": "AlphaZero", "level": "model", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "reinforcement learning", "modalities_in": [], "modalities_out": [ "action" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-12-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": "model_class left empty: AURION has no reinforcement learning class yet.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "General reinforcement learning algorithm that mastered chess and shogi by self-play from the rules alone. Paper by Silver et al. (2017).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1712.01815" ], "unverified": [ "developer" ] }, { "id": "google-deepmind-muzero", "name": "MuZero", "level": "model", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "reinforcement learning", "modalities_in": [], "modalities_out": [ "action" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-11-19", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": "model_class left empty: AURION has no reinforcement learning class yet.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Agent that plans with a learned model and mastered Atari, Go, chess and shogi without being given the rules. Paper by Schrittwieser et al. (2019).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1911.08265" ], "unverified": [ "developer" ] }, { "id": "google-deepmind-agent57", "name": "Agent57", "level": "model", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "reinforcement learning", "modalities_in": [], "modalities_out": [ "action" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-03-30", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": "model_class left empty: AURION has no reinforcement learning class yet.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "Deep reinforcement learning agent that outperforms the standard human benchmark on Atari games. Paper by Badia et al. (2020).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2003.13350" ], "unverified": [ "developer" ] }, { "id": "google-deepmind-alphastar", "name": "AlphaStar", "level": "model", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "reinforcement learning", "modalities_in": [], "modalities_out": [ "action" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-01-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": "model_class left empty: AURION has no reinforcement learning class yet.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "DeepMind agent that plays the real-time strategy game StarCraft II.", "verified_on": "2026-10-12", "sources": [ "https://deepmind.google/discover/blog/alphastar-mastering-the-real-time-strategy-game-starcraft-ii/" ], "unverified": [] }, { "id": "google-deepmind-alphadev", "name": "AlphaDev", "level": "model", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "reinforcement learning", "modalities_in": [], "modalities_out": [ "action" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-06-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": "model_class left empty: AURION has no reinforcement learning class yet. The sorting algorithms it found were open-sourced in LLVM libc++ (coverage ledger).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "DeepMind reinforcement learning agent that discovered faster sorting algorithms at the assembly level.", "verified_on": "2026-10-12", "sources": [ "https://deepmind.google/discover/blog/alphadev-discovers-faster-sorting-algorithms/" ], "unverified": [] }, { "id": "google-deepmind-alphatensor", "name": "AlphaTensor", "level": "model", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "reinforcement learning", "modalities_in": [], "modalities_out": [ "action" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-10-05", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "model_class left empty: AURION has no reinforcement learning class yet. Licence is that of the google-deepmind/alphatensor repository (Apache-2.0).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "DeepMind reinforcement learning agent that discovered new, provably correct algorithms for matrix multiplication.", "verified_on": "2026-10-12", "sources": [ "https://deepmind.google/discover/blog/discovering-novel-algorithms-with-alphatensor/", "https://github.com/google-deepmind/alphatensor" ], "unverified": [] }, { "id": "openai-five", "name": "OpenAI Five", "level": "model", "family_id": null, "developer": [ "openai" ], "model_class": "reinforcement learning", "modalities_in": [], "modalities_out": [ "action" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2019-12-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": "model_class left empty: AURION has no reinforcement learning class yet.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": null, "summary": "OpenAI agent that became the first AI system to defeat the world champions at the esports game Dota 2, in April 2019. Paper by OpenAI et al. (2019).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1912.06680" ], "unverified": [] }, { "id": "google-txgemma", "name": "TxGemma", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "scientific", "modalities_in": [ "text", "molecule" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Published sizes include 2B and 9B predict models and a 27B chat model (Hugging Face).", "context_length": null, "release_date": "2025-03-25", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": "Gated on Hugging Face with licence tag \"other\"; the terms page was not opened.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google suite of efficient, generalist language models for therapeutic property prediction, with interactive reasoning and explanation in the chat versions. Paper by Wang et al. (2025).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2504.06196", "https://huggingface.co/google/txgemma-9b-predict", "https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/google-models", "https://developers.googleblog.com/en/introducing-txgemma-open-models-improving-therapeutics-development/" ], "unverified": [ "openness" ] }, { "id": "google-txgemma-2b-predict", "name": "TxGemma 2B predict", "level": "variant", "family_id": "google-txgemma", "developer": [ "google" ], "model_class": "scientific", "modalities_in": [ "text", "molecule" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03-21", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "TxGemma 2B predict checkpoint of TxGemma for therapeutic property prediction.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/google/txgemma-2b-predict", "https://arxiv.org/abs/2504.06196" ], "unverified": [ "openness" ] }, { "id": "google-txgemma-9b-predict", "name": "TxGemma 9B predict", "level": "variant", "family_id": "google-txgemma", "developer": [ "google" ], "model_class": "scientific", "modalities_in": [ "text", "molecule" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03-21", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "TxGemma 9B predict checkpoint of TxGemma for therapeutic property prediction.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/google/txgemma-9b-predict", "https://arxiv.org/abs/2504.06196" ], "unverified": [ "openness" ] }, { "id": "google-txgemma-27b-chat", "name": "TxGemma 27B chat", "level": "variant", "family_id": "google-txgemma", "developer": [ "google" ], "model_class": "scientific", "modalities_in": [ "text", "molecule" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03-21", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "TxGemma 27B chat checkpoint of TxGemma for therapeutic property prediction and conversation.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/google/txgemma-27b-chat", "https://arxiv.org/abs/2504.06196" ], "unverified": [ "openness" ] }, { "id": "c2s-scale", "name": "C2S-Scale (Cell2Sentence)", "level": "family", "family_id": null, "developer": [ "yale-university", "google", "google-deepmind" ], "model_class": "scientific", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": "Gemma 2 models trained on single-cell data written as text (cell sentences)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "2B and 27B checkpoints on Hugging Face; also on Kaggle Models and KerasHub.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Language models for single-cell analysis that read gene expression profiles written as sentences, from a collaboration of Yale University, Google Research and Google DeepMind.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/vandijklab/C2S-Scale-Gemma-2-27B", "https://github.com/vandijklab/cell2sentence", "https://www.kaggle.com/models/keras/cell2sentence", "https://ai.azure.com/catalog/models?industry=health-and-life-sciences" ], "unverified": [] }, { "id": "c2s-scale-gemma-2-2b", "name": "C2S-Scale-Gemma-2-2B", "level": "variant", "family_id": "c2s-scale", "developer": [ "yale-university", "google", "google-deepmind" ], "model_class": "scientific", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10-06", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "2B C2S-Scale model, fine-tuned from Gemma 2 2B for single-cell analysis.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/vandijklab/C2S-Scale-Gemma-2-2B" ], "unverified": [] }, { "id": "c2s-scale-gemma-2-27b", "name": "C2S-Scale-Gemma-2-27B", "level": "variant", "family_id": "c2s-scale", "developer": [ "yale-university", "google", "google-deepmind" ], "model_class": "scientific", "modalities_in": [ "text" ], "modalities_out": [ "text" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10-06", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "27B C2S-Scale model, fine-tuned from Gemma 2 27B for single-cell analysis.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/vandijklab/C2S-Scale-Gemma-2-27B" ], "unverified": [] }, { "id": "geneformer", "name": "Geneformer", "level": "model", "family_id": null, "developer": [], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "dense transformer", "architecture_detail": "Encoder pre-trained on single-cell transcriptomes (fill-mask)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Original model trained on about 30 million transcriptomes; expanded model on about 104 million (model card). NVIDIA BioNeMo publishes 10M and 100M builds.", "context_length": null, "release_date": "2022-03-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Transformer pre-trained on human single-cell transcriptomes for predictions in network biology where data are limited, such as in silico perturbation and cell classification.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/ctheodoris/Geneformer", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/clara/models/geneformer_100m" ], "unverified": [] }, { "id": "metagene-1", "name": "METAGENE-1", "level": "model", "family_id": null, "developer": [ "usc", "prime-intellect" ], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "autoregressive", "architecture_detail": "7B autoregressive transformer on metagenomic sequences", "params_total": 7000000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-01-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Metagenomic foundation model trained on over 1.5T base pairs of DNA and RNA sequenced from wastewater, for pandemic monitoring. Paper by Liu et al. (2025).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2501.02045", "https://huggingface.co/metagene-ai/METAGENE-1", "https://www.kaggle.com/models/metagene/metagene-1", "https://arxiv.org/html/2501.02045" ], "unverified": [] }, { "id": "recursion-openphenom", "name": "OpenPhenom-S/16", "level": "model", "family_id": null, "developer": [ "recursion" ], "model_class": "scientific", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": "Channel-agnostic masked autoencoder with a ViT-S/16 encoder", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-10-21", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Recursion image encoder for featurising cell microscopy images, trained on RxRx3 and JUMP-CP data.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/recursionpharma/OpenPhenom", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden" ], "unverified": [] }, { "id": "microsoft-bioemu", "name": "BioEmu", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "other", "architecture_detail": null, "params_total": 31000000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-02-19", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft Biomolecular Emulator: given a protein sequence, it samples equilibrium structure ensembles far faster than molecular dynamics.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/microsoft/bioemu", "https://github.com/microsoft/bioemu", "https://ai.azure.com/catalog/models?industry=health-and-life-sciences" ], "unverified": [ "summary" ] }, { "id": "microsoft-evodiff", "name": "EvoDiff", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "diffusion", "architecture_detail": "Discrete diffusion over protein sequences and evolutionary alignments", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft generative model for protein sequences and multiple sequence alignments based on discrete diffusion.", "verified_on": "2026-10-12", "sources": [ "https://github.com/microsoft/evodiff", "https://ai.azure.com/catalog/models?industry=health-and-life-sciences" ], "unverified": [] }, { "id": "microsoft-dayhoff-3b-gr-hm-c", "name": "Dayhoff-3b-GR-HM-c", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "autoregressive", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Part of the Dayhoff Atlas of protein data and models (model card).", "context_length": null, "release_date": "2025-07-04", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft protein language model from the Dayhoff Atlas, trained on metagenomic, natural and structure-derived protein sequences; predicts mutation effects and generates proteins.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/microsoft/Dayhoff-3b-GR-HM-c", "https://ai.azure.com/catalog/models?industry=health-and-life-sciences" ], "unverified": [] }, { "id": "proteinmpnn", "name": "ProteinMPNN", "level": "model", "family_id": null, "developer": [], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "GNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Message-passing neural network that designs amino acid sequences for a given protein backbone.", "verified_on": "2026-10-12", "sources": [ "https://github.com/dauparas/ProteinMPNN", "https://ai.azure.com/catalog/models?industry=health-and-life-sciences" ], "unverified": [ "summary" ] }, { "id": "rfdiffusion", "name": "RFdiffusion", "level": "model", "family_id": null, "developer": [], "model_class": "scientific", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": "Repository RosettaCommons/RFdiffusion; GitHub reports no standard licence identifier.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Diffusion model for designing new protein structures, run from the RosettaCommons/RFdiffusion code.", "verified_on": "2026-10-12", "sources": [ "https://github.com/RosettaCommons/RFdiffusion", "https://ai.azure.com/catalog/models?industry=health-and-life-sciences" ], "unverified": [ "summary" ] }, { "id": "lfads", "name": "LFADS", "level": "family", "family_id": null, "developer": [ "google", "stanford-university" ], "model_class": "scientific", "modalities_in": [ "time series" ], "modalities_out": [ "time series" ], "architecture": "other", "architecture_detail": "Sequential variational autoencoder", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2016-08-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Latent Factor Analysis via Dynamical Systems, a method for inferring latent dynamics from neural population recordings. Paper by Sussillo et al. (2016); listed in TensorFlow Model Garden research.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1608.06315", "https://github.com/tensorflow/models/blob/master/research/README.md", "https://arxiv.org/html/1608.06315" ], "unverified": [] }, { "id": "marco-crystallization", "name": "MARCO crystallisation classifier", "level": "family", "family_id": null, "developer": [ "google" ], "model_class": "scientific", "modalities_in": [ "image" ], "modalities_out": [ "text" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-03-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Deep convolutional network that classifies protein crystallisation trial images. Paper by Bruno et al. (2018); listed in TensorFlow Model Garden research.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1803.10342", "https://github.com/tensorflow/models/blob/master/research/README.md", "https://arxiv.org/html/1803.10342" ], "unverified": [] }, { "id": "microsoft-mattergen", "name": "MatterGen", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "materials", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "diffusion", "architecture_detail": "Diffusion over atom coordinates, elements and lattice", "params_total": 46800000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-12-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft Research generative model for inorganic materials design that can be fine-tuned to generate crystals with target properties.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2312.03687", "https://huggingface.co/microsoft/mattergen", "https://github.com/microsoft/mattergen", "https://ai.azure.com/catalog/models?task=materials-design" ], "unverified": [] }, { "id": "microsoft-mattersim", "name": "MatterSim", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "materials", "modalities_in": [ "molecule" ], "modalities_out": [ "molecule" ], "architecture": "other", "architecture_detail": "Deep learning atomistic model (machine-learned interatomic potential)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-05-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft deep learning atomistic model that simulates materials across elements, temperatures and pressures. Paper by Yang et al. (2024).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2405.04967", "https://github.com/microsoft/mattersim", "https://ai.azure.com/catalog/models?task=materials-design" ], "unverified": [] }, { "id": "nvidia-fourcastnet", "name": "FourCastNet", "level": "family", "family_id": null, "developer": [ "nvidia" ], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "other", "architecture_detail": "Adaptive Fourier neural operator (first version); later spherical Fourier neural operators", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-02-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "open source", "licence": "apache-2-0", "access_routes": [], "deployment_tiers": [], "footprint_note": "Licence is that of NVIDIA PhysicsNeMo (Apache-2.0), which hosts the models.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Global data-driven high-resolution weather model family from NVIDIA, first built on adaptive Fourier neural operators. Paper by Pathak et al. (2022).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2202.11214", "https://github.com/NVIDIA/physicsnemo", "https://catalog.ngc.nvidia.com/orgs/nim/teams/nvidia/models/earth2-sfno-era5-73ch" ], "unverified": [] }, { "id": "nvidia-fourcastnet-sfno", "name": "FourCastNet SFNO", "level": "variant", "family_id": "nvidia-fourcastnet", "developer": [ "nvidia" ], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "other", "architecture_detail": "Spherical Fourier neural operator", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-06-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "FourCastNet version built on spherical Fourier neural operators for stable long rollouts on the sphere; NGC publishes an ERA5 73-channel checkpoint. Paper by Bonev et al. (2023).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2306.03838", "https://catalog.ngc.nvidia.com/orgs/nim/teams/nvidia/models/earth2-sfno-era5-73ch" ], "unverified": [] }, { "id": "nvidia-fourcastnet-3", "name": "FourCastNet 3", "level": "variant", "family_id": "nvidia-fourcastnet", "developer": [ "nvidia" ], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "other", "architecture_detail": "Geometric probabilistic ensemble forecasting", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-07-16", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Probabilistic ensemble global weather model that NVIDIA reports as 8 to 60 times faster than comparable approaches. Paper by Bonev et al. (2025).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2507.12144", "https://huggingface.co/nvidia/fourcastnet3", "https://ai.azure.com/catalog/models?task=weather-forecasting" ], "unverified": [] }, { "id": "nvidia-corrdiff", "name": "CorrDiff", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "diffusion", "architecture_detail": "Regression plus residual corrective diffusion", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-09-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Generative downscaling model that turns coarse weather data into kilometre-scale fields, part of NVIDIA Earth-2. Paper by Mardani et al. (2023).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2309.15214", "https://catalog.ngc.nvidia.com/orgs/nim/teams/nvidia/models/earth2-corrdiff-us-gefs-hrrr" ], "unverified": [] }, { "id": "nvidia-stormcast", "name": "StormCast", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-08-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Generative diffusion model that emulates kilometre-scale convection-allowing weather models, part of NVIDIA Earth-2. Paper by Pathak et al. (2024).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2408.10958", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/earth-2/models/stormcast-v1-era5-hrrr" ], "unverified": [] }, { "id": "nvidia-cbottle", "name": "Climate in a Bottle (cBottle)", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-05-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Generative foundation model for the kilometre-scale global atmosphere, part of NVIDIA Earth-2. Paper by Brenowitz et al. (2025).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2505.06474", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/earth-2/models/cbottle" ], "unverified": [] }, { "id": "dlesym", "name": "DLESyM", "level": "model", "family_id": null, "developer": [ "university-of-washington" ], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-09-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Parsimonious deep learning Earth system model that simulates the current climate over 1000-year periods without drift. Paper by Cresswell-Clay et al. (2024).", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2409.16247", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/earth-2/models/dlesym-v1-era5" ], "unverified": [ "developer" ] }, { "id": "meteoswiss-varda-single-1-0", "name": "Varda-single-1.0", "level": "model", "family_id": null, "developer": [ "meteoswiss" ], "model_class": "weather", "modalities_in": [ "sensor" ], "modalities_out": [ "sensor" ], "architecture": "other", "architecture_detail": "Built with the Anemoi framework (model card); Hugging Face task graph-ml", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-10-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "cc-by-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MeteoSwiss data-driven medium-range weather forecasting system for the Alpine domain: hourly forecasts on a 1 km mesh over Switzerland plus 31 km global forecasts.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2610.01835", "https://huggingface.co/MeteoSwiss/Varda-single-1.0", "https://huggingface.co/api/models?pipeline_tag=graph-ml&sort=likes&limit=50" ], "unverified": [] }, { "id": "edge-impulse-fomo", "name": "FOMO (Faster Objects, More Objects)", "level": "model", "family_id": null, "developer": [ "edge-impulse" ], "model_class": "TinyML", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": "MobileNetV2 trunk; outputs object centroids instead of bounding boxes", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "microcontroller", "embedded GPU" ], "footprint_note": "Edge Impulse states 30 fps on an Arduino Nicla Vision (Cortex-M7) using 245K RAM, about 10 fps under 100KB RAM on a Cortex-M4F at 80MHz, and 60 fps on a Raspberry Pi 4; minimum 256 KB RAM for 64x64 images.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Edge Impulse object detection learning block for constrained devices that predicts object centroids rather than boxes. Used for counting and locating objects on microcontrollers.", "verified_on": "2026-10-12", "sources": [ "https://docs.edgeimpulse.com/studio/projects/learning-blocks/blocks/object-detection/fomo" ], "unverified": [] }, { "id": "edge-impulse-fomo-ad", "name": "FOMO-AD (Visual Anomaly Detection)", "level": "model", "family_id": null, "developer": [ "edge-impulse" ], "model_class": "TinyML", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "other", "architecture_detail": "Selectable backbone with PatchCore or Gaussian Mixture Model anomaly scoring", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": "The page says GMM scoring is the default for a low-power device and PatchCore for a high-power device.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Edge Impulse visual anomaly detection block, previously named FOMO-AD, that flags image regions that do not match normal training images. Meant for defect detection when few defect samples exist.", "verified_on": "2026-10-12", "sources": [ "https://docs.edgeimpulse.com/studio/projects/learning-blocks/blocks/visual-anomaly-detection-fomo-ad" ], "unverified": [] }, { "id": "google-tflm-micro-speech", "name": "Micro Speech (TFLM keyword model)", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "TinyML", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Repository licence Apache-2.0 per the coverage entry.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open source", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "microcontroller" ], "footprint_note": "Under 20 kB model that recognises two keywords, yes and no, from spectrogram features (TFLM README).", "runtimes": [ "tflm" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "TensorFlow Lite Micro example wake-word model that recognises the keywords yes and no. It is the reference keyword-spotting example for microcontrollers.", "verified_on": "2026-10-12", "sources": [ "https://github.com/tensorflow/tflite-micro/tree/main/tensorflow/lite/micro/examples/micro_speech" ], "unverified": [] }, { "id": "dtln", "name": "DTLN (Dual-signal Transformation LSTM Network)", "level": "model", "family_id": null, "developer": [], "model_class": "TinyML", "modalities_in": [ "audio" ], "modalities_out": [ "audio" ], "architecture": "other", "architecture_detail": "Stacked LSTM network combining STFT and a learned analysis and synthesis basis; under one million parameters", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-05-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": "TFLM example runs the 8-bit quantised noise suppression part, retrained by Cadence, on a HiFi DSP (TFLM dtln README).", "runtimes": [ "tflm" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time speech noise suppression network from the Deep Noise Suppression Challenge. A quantised version is a TensorFlow Lite Micro example for DSPs.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2005.07551", "https://github.com/tensorflow/tflite-micro/tree/main/tensorflow/lite/micro/examples/dtln" ], "unverified": [] }, { "id": "arm-hello-edge-kws", "name": "Hello Edge keyword spotting models", "level": "family", "family_id": null, "developer": [ "arm" ], "model_class": "TinyML", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Arm ML-Zoo ships CNN, DNN and DS-CNN keyword models in small, medium and large sizes (INT8 and FP32 TFLite). DS-CNN is held as its own AURION record (ds-cnn).", "context_length": null, "release_date": "2017-11-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "microcontroller" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Keyword spotting network families compared for microcontrollers in the Hello Edge study. Arm publishes the trained models in its ML-Zoo for Cortex-M and Ethos-U targets.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1711.07128", "https://github.com/ARM-software/ML-zoo/tree/master/models/keyword_spotting" ], "unverified": [] }, { "id": "arm-hello-edge-kws-cnn", "name": "Hello Edge KWS CNN (small, medium, large)", "level": "variant", "family_id": "arm-hello-edge-kws", "developer": [ "arm" ], "model_class": "TinyML", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Small, medium and large sizes in INT8 and FP32 TFLite; Arm ML-Zoo README ticks Cortex-M support. Repository licence Apache-2.0 per the coverage entry.", "context_length": null, "release_date": "2017-11-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "microcontroller" ], "footprint_note": null, "runtimes": [ "tflm" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "CNN keyword spotting model from the Hello Edge study, published in Arm ML-Zoo for microcontroller deployment.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1711.07128", "https://github.com/ARM-software/ML-zoo", "https://github.com/ARM-software/ML-zoo/tree/master/models/keyword_spotting" ], "unverified": [] }, { "id": "arm-hello-edge-kws-dnn", "name": "Hello Edge KWS DNN (small, medium, large)", "level": "variant", "family_id": "arm-hello-edge-kws", "developer": [ "arm" ], "model_class": "TinyML", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Small, medium and large sizes in INT8 and FP32 TFLite; Arm ML-Zoo README ticks Cortex-M support. Repository licence Apache-2.0 per the coverage entry.", "context_length": null, "release_date": "2017-11-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "microcontroller" ], "footprint_note": null, "runtimes": [ "tflm" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "DNN keyword spotting model from the Hello Edge study, published in Arm ML-Zoo for microcontroller deployment.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1711.07128", "https://github.com/ARM-software/ML-zoo", "https://github.com/ARM-software/ML-zoo/tree/master/models/keyword_spotting" ], "unverified": [] }, { "id": "stmicroelectronics-arc-fault-detection", "name": "ST arc fault detection models (st_conv, st_dense)", "level": "model", "family_id": null, "developer": [ "stmicroelectronics" ], "model_class": "TinyML", "modalities_in": [ "sensor" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "st_conv is a per-channel CNN (1 or 4 channels, 512-sample windows); st_dense is a fully connected network (1 or 8 channels).", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "microcontroller" ], "footprint_note": "Benchmarked in the README on the B-U585I-IOT02A board (STM32U5) and STM32N6.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "STMicroelectronics models that classify electrical current windows as normal or arc fault. Built for arc fault detection on STM32 microcontrollers.", "verified_on": "2026-10-12", "sources": [ "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/arc_fault_detection", "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/arc_fault_detection/st_conv", "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/arc_fault_detection/st_dense" ], "unverified": [] }, { "id": "stmicroelectronics-miniresnet-aed", "name": "ST MiniResNet v1 and v2 audio event detection", "level": "model", "family_id": null, "developer": [ "stmicroelectronics" ], "model_class": "TinyML", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Resized ResNet18 (v1) and ResNet18v2 (v2) with a configurable number of stacks, run on mel-spectrograms.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "microcontroller" ], "footprint_note": "Benchmarked in the README on NUCLEO-U3C5ZI and B-U585I-IOT02A (STM32U5) boards.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Quantised small ResNet models for audio event detection from mel-spectrograms. Part of the STM32 model zoo for microcontrollers.", "verified_on": "2026-10-12", "sources": [ "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/audio_event_detection", "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/audio_event_detection/miniresnetv1", "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/audio_event_detection/miniresnetv2" ], "unverified": [] }, { "id": "stmicroelectronics-cnn2d-hand-posture", "name": "ST CNN2D hand posture (Time-of-Flight)", "level": "model", "family_id": null, "developer": [ "stmicroelectronics" ], "model_class": "TinyML", "modalities_in": [ "sensor" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Recommended input 8 x 8 x 2 (distance and signal) from an ST multi-zone Time-of-Flight sensor.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "microcontroller" ], "footprint_note": "README lists STM32F4, STM32F401, STM32L4 and STM32U5 targets.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ST-designed CNN that recognises static hand postures from multi-zone Time-of-Flight sensor data. Used for touchless gesture control on STM32 microcontrollers.", "verified_on": "2026-10-12", "sources": [ "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/hand_posture", "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/hand_posture/st_cnn2d_handposture" ], "unverified": [] }, { "id": "stmicroelectronics-har", "name": "ST human activity recognition models (st_gmp, st_ign)", "level": "model", "family_id": null, "developer": [ "stmicroelectronics" ], "model_class": "TinyML", "modalities_in": [ "sensor" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "st_gmp uses global max pooling; st_ign is a modified Ignatov CNN. Window lengths 24 and 48 samples; trained on WISDM and an ST mobility dataset.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "microcontroller" ], "footprint_note": "README benchmarks on B-U585I-IOT02A (STM32U5) and STM32L4.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "CNN models that classify human activities from 3-axis accelerometer windows. Built for wearables on STM32 microcontrollers.", "verified_on": "2026-10-12", "sources": [ "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/human_activity_recognition", "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/human_activity_recognition/st_gmp", "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/human_activity_recognition/st_ign" ], "unverified": [] }, { "id": "stmicroelectronics-st-mnist-v1", "name": "ST MNIST v1", "level": "model", "family_id": null, "developer": [ "stmicroelectronics" ], "model_class": "TinyML", "modalities_in": [ "image" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Depthwise separable CNN for 28 x 28 inputs, int8 quantised.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "microcontroller" ], "footprint_note": "README lists STM32H7, STM32U5, STM32N6 and STM32MP2 targets.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small depthwise separable CNN for MNIST-type tasks such as digit, letter or fashion item recognition on STM32 devices.", "verified_on": "2026-10-12", "sources": [ "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/image_classification/st_mnistv1" ], "unverified": [] }, { "id": "stmicroelectronics-st-yolo", "name": "ST YOLO detectors", "level": "family", "family_id": null, "developer": [ "stmicroelectronics" ], "model_class": "TinyML", "modalities_in": [ "image" ], "modalities_out": [], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "ST YOLO LC v1 and ST YOLOX-nano (TensorFlow) plus ST YOLOD v2 Tiny and Milli (PyTorch) in the STM32 model zoo.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "microcontroller" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Object detectors optimised by STMicroelectronics for real-time detection on STM32 devices, including the STM32N6 with its neural processing unit.", "verified_on": "2026-10-12", "sources": [ "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/object_detection" ], "unverified": [] }, { "id": "stmicroelectronics-st-yolo-lc-v1", "name": "ST YOLO LC v1", "level": "variant", "family_id": "stmicroelectronics-st-yolo", "developer": [ "stmicroelectronics" ], "model_class": "TinyML", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "microcontroller" ], "footprint_note": "README benchmarks on STM32N6570-DK, STM32H7 and STM32MP257F-DK2.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time int8 object detection model from STMicroelectronics, implemented in TensorFlow for STM32 targets.", "verified_on": "2026-10-12", "sources": [ "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/object_detection/st_yololcv1" ], "unverified": [] }, { "id": "stmicroelectronics-st-yolox-nano", "name": "ST YOLOX-nano", "level": "variant", "family_id": "stmicroelectronics-st-yolo", "developer": [ "stmicroelectronics" ], "model_class": "TinyML", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "microcontroller" ], "footprint_note": "README benchmarks on STM32N6570-DK and STM32H747I-DISCO.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "STMicroelectronics optimised version of YOLOX quantised to int8 for real-time detection on STM32 devices.", "verified_on": "2026-10-12", "sources": [ "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/object_detection/st_yoloxn" ], "unverified": [] }, { "id": "stmicroelectronics-st-yolod-v2", "name": "ST YOLOD v2 (Tiny, Milli)", "level": "variant", "family_id": "stmicroelectronics-st-yolo", "developer": [ "stmicroelectronics" ], "model_class": "TinyML", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "microcontroller" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "PyTorch YOLO detectors from STMicroelectronics in Tiny and Milli sizes, listed in the STM32 model zoo object detection section.", "verified_on": "2026-10-12", "sources": [ "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/object_detection" ], "unverified": [] }, { "id": "stmicroelectronics-stft-tcnn", "name": "STFT-TCNN speech enhancement", "level": "model", "family_id": null, "developer": [ "stmicroelectronics" ], "model_class": "TinyML", "modalities_in": [ "audio" ], "modalities_out": [ "audio" ], "architecture": "CNN", "architecture_detail": "TCN core of Pandey and Wang TCNN with STFT pre-processing and inverse STFT post-processing so it quantises to int8", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "microcontroller" ], "footprint_note": "README benchmarks on STM32N6 (STM32N6570-DK).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Frequency-domain temporal convolutional network that masks noise in speech spectrograms. Adapted by STMicroelectronics for int8 use on STM32N6.", "verified_on": "2026-10-12", "sources": [ "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/speech_enhancement", "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/speech_enhancement/stft_tcnn" ], "unverified": [] }, { "id": "stmicroelectronics-xinet-picasso-muse", "name": "XiNet Picasso Muse style transfer", "level": "model", "family_id": null, "developer": [ "stmicroelectronics" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": "XiNet backbone for tinyML, int8 quantised", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "microcontroller" ], "footprint_note": "README lists STM32N6 (STM32N6570-DK), STM32U5 and STM32MP2.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Lightweight neural style transfer model built on XiNets, networks designed for microcontrollers. Runs on STM32 devices.", "verified_on": "2026-10-12", "sources": [ "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/neural_style_transfer", "https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/neural_style_transfer/xinet_picasso_muse" ], "unverified": [] }, { "id": "silicon-labs-mltk-kws", "name": "Silicon Labs MLTK keyword spotting models", "level": "family", "family_id": null, "developer": [ "silicon-labs" ], "model_class": "TinyML", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "MLTK reference models: Keyword Spotting On/Off (v1 to v3), MobileNetv2, with Transfer Learning, Pac-Man (v1 to v3), Alexa and Numbers.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "microcontroller" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Keyword and wake-word models that Silicon Labs publishes with its Machine Learning Toolkit tutorials for EFR32 wireless microcontrollers.", "verified_on": "2026-10-12", "sources": [ "https://siliconlabs.github.io/mltk/docs/python_api/models/siliconlabs/index.html", "https://github.com/SiliconLabs/mltk/tree/master/mltk/models/siliconlabs" ], "unverified": [] }, { "id": "silicon-labs-kws-on-off-v3", "name": "Keyword Spotting On/Off v3", "level": "variant", "family_id": "silicon-labs-mltk-kws", "developer": [ "silicon-labs" ], "model_class": "TinyML", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": "TENet (temporal and depthwise convolutions), 107 layers, about 6.1M MACs", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "microcontroller" ], "footprint_note": "Profiled on an EFR32xG24 with the MVP accelerator at 78 MHz: 84.8k RAM, 531.7k model file, about 75 ms per inference.", "runtimes": [ "tflm" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "CNN classifier that detects the keywords on and off plus an unknown class. A Silicon Labs MLTK tutorial model for EFR32 microcontrollers.", "verified_on": "2026-10-12", "sources": [ "https://siliconlabs.github.io/mltk/docs/python_api/models/siliconlabs/keyword_spotting_on_off_v3.html" ], "unverified": [] }, { "id": "silicon-labs-kws-numbers", "name": "Keyword Spotting Numbers", "level": "variant", "family_id": "silicon-labs-mltk-kws", "developer": [ "silicon-labs" ], "model_class": "TinyML", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": "CNN plus LSTM", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "CNN and LSTM model that detects the spoken words zero to nine, built for the MLTK quantised LSTM tutorial.", "verified_on": "2026-10-12", "sources": [ "https://siliconlabs.github.io/mltk/docs/python_api/models/siliconlabs/keyword_spotting_numbers.html", "https://siliconlabs.github.io/mltk/docs/python_api/models/siliconlabs/index.html" ], "unverified": [] }, { "id": "silicon-labs-fingerprint-signature-generator", "name": "Fingerprint Signature Generator", "level": "model", "family_id": null, "developer": [ "silicon-labs" ], "model_class": "TinyML", "modalities_in": [ "image" ], "modalities_out": [], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Silicon Labs MLTK model built for the fingerprint authentication tutorial; it turns a fingerprint image into a signature for matching.", "verified_on": "2026-10-12", "sources": [ "https://siliconlabs.github.io/mltk/docs/python_api/models/siliconlabs/fingerprint_signature_generator.html", "https://siliconlabs.github.io/mltk/docs/python_api/models/siliconlabs/index.html" ], "unverified": [] }, { "id": "qualcomm-bc-resnet", "name": "BC-ResNet (Broadcasted Residual Network)", "level": "model", "family_id": null, "developer": [ "qualcomm" ], "model_class": "TinyML", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-06-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "The paper targets devices with limited resources such as mobile phones; Silicon Labs MLTK ships a kws_bc_resnet port.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Keyword spotting network that mixes 1D temporal and 2D convolutions through broadcasted residual connections to keep accuracy high at small size.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2106.04140", "https://github.com/SiliconLabs/mltk/tree/master/mltk/models/siliconlabs" ], "unverified": [] }, { "id": "tenet-kws", "name": "TENet (Temporal Efficient Neural Network)", "level": "model", "family_id": null, "developer": [], "model_class": "TinyML", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-10-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [ "microcontroller" ], "footprint_note": "Silicon Labs Keyword Spotting On/Off v3 is based on TENet and runs on an EFR32xG24 microcontroller.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small-footprint keyword spotting network using temporal and depthwise convolutions with multi-scale temporal convolution blocks.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2010.09960", "https://github.com/SiliconLabs/mltk/tree/master/mltk/models/shared", "https://siliconlabs.github.io/mltk/docs/python_api/models/siliconlabs/keyword_spotting_on_off_v3.html" ], "unverified": [] }, { "id": "nxp-microspeech-lstm", "name": "Microspeech LSTM", "level": "model", "family_id": null, "developer": [ "nxp" ], "model_class": "TinyML", "modalities_in": [ "audio" ], "modalities_out": [ "labels and boxes" ], "architecture": "other", "architecture_detail": "RNN (LSTM)", "params_total": 119923, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "About 120K parameters, 122 KB FP32 or 123 KB INT8; listed as compatible with i.MX 8MP and i.MX 93.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "LSTM command recognition model trained on the Speech Commands dataset, published in the NXP eIQ model zoo.", "verified_on": "2026-10-12", "sources": [ "https://github.com/NXP/eiq-model-zoo/tree/main/tasks/audio/command-recognition" ], "unverified": [] }, { "id": "eeg-tcnet", "name": "EEG-TCNet", "level": "model", "family_id": null, "developer": [ "eth-zurich" ], "model_class": "TinyML", "modalities_in": [ "sensor" ], "modalities_out": [ "labels and boxes" ], "architecture": "CNN", "architecture_detail": "Temporal convolutional network", "params_total": 4096, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2020-05-31", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "NXP eIQ lists 4,096 parameters, 28 KB INT8, compatible with i.MX 8MP and i.MX 93. The paper targets embedded classification on resource-limited edge devices.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Temporal convolutional network that classifies motor-imagery EEG signals for brain-machine interfaces with very few parameters.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2006.00622", "https://github.com/NXP/eiq-model-zoo/tree/main/tasks/misc" ], "unverified": [] }, { "id": "fast-srgan", "name": "Fast-SRGAN", "level": "model", "family_id": null, "developer": [], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": 168000, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": "NXP eIQ lists 168K parameters, 240 KB INT8, compatible with i.MX 8M Plus and i.MX 93.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Small SRGAN-style super-resolution model trained on DIV2K, published as an NXP eIQ model zoo recipe for real-time upscaling.", "verified_on": "2026-10-12", "sources": [ "https://github.com/NXP/eiq-model-zoo/tree/main/tasks/vision/super-resolution" ], "unverified": [] }, { "id": "black-forest-labs-flux-1-schnell", "name": "FLUX.1 [schnell]", "level": "variant", "family_id": "black-forest-labs-flux-1", "developer": [ "black-forest-labs" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "flow matching", "architecture_detail": "Rectified flow transformer distilled to 1 to 4 steps", "params_total": 12000000000, "params_active": null, "params_checkpoint": null, "params_note": "12 billion parameters (model card).", "context_length": null, "release_date": "2024-07-31", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fastest FLUX.1 text-to-image model, a 12B rectified flow transformer that generates images in 1 to 4 steps under Apache 2.0.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/black-forest-labs/FLUX.1-schnell", "https://replicate.com/black-forest-labs/flux-schnell" ], "unverified": [] }, { "id": "black-forest-labs-flux-1-tools", "name": "FLUX.1 Tools (Fill, Depth, Canny, Redux)", "level": "model", "family_id": "black-forest-labs-flux-1", "developer": [ "black-forest-labs" ], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": "flow matching", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Open [dev] weights for Fill, Depth, Canny and Redux under the FLUX.1 dev licence; Fill, Depth and Canny also as [pro] API models. Depth and Canny are deprecated in the API.", "context_length": null, "release_date": "2024-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": "flux-non-commercial", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Add-on FLUX.1 models for inpainting and outpainting (Fill), structure guidance from depth maps or edges (Depth, Canny) and image variation (Redux).", "verified_on": "2026-10-12", "sources": [ "https://bfl.ai/announcements/24-11-21-tools", "https://huggingface.co/black-forest-labs/FLUX.1-Fill-dev", "https://replicate.com/black-forest-labs/flux-fill-pro" ], "unverified": [] }, { "id": "black-forest-labs-flux-2-pro", "name": "FLUX.2 [pro]", "level": "model", "family_id": "black-forest-labs-flux-2", "developer": [ "black-forest-labs" ], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-11-25", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "High-quality FLUX.2 image generation and editing with up to eight reference images. Hosted only through the BFL API, playground and partners.", "verified_on": "2026-10-12", "sources": [ "https://bfl.ai/blog/flux-2", "https://replicate.com/black-forest-labs/flux-2-pro" ], "unverified": [] }, { "id": "black-forest-labs-flux-2-flex", "name": "FLUX.2 [flex]", "level": "model", "family_id": "black-forest-labs-flux-2", "developer": [ "black-forest-labs" ], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-11-25", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "FLUX.2 variant that exposes steps and guidance scale so developers can trade quality and speed; supports up to ten reference images. Hosted only through the BFL API, playground and partners.", "verified_on": "2026-10-12", "sources": [ "https://bfl.ai/blog/flux-2", "https://replicate.com/black-forest-labs/flux-2-flex" ], "unverified": [] }, { "id": "stability-ai-sd-turbo", "name": "SD-Turbo", "level": "model", "family_id": "stability-ai-stable-diffusion", "developer": [ "stability-ai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": 865910724, "params_active": null, "params_checkpoint": null, "params_note": "SDXL-Turbo is the larger sibling (2.57B safetensors total, Stability AI non-commercial community licence). Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2023-11-27", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Distilled Stable Diffusion model that generates an image in one or a few steps. Released by Stability AI on Hugging Face.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/stabilityai/sd-turbo", "https://huggingface.co/api/models?pipeline_tag=text-to-image&sort=downloads&limit=50", "https://huggingface.co/stabilityai/sdxl-turbo" ], "unverified": [] }, { "id": "stability-ai-stable-image-ultra", "name": "Stable Image Ultra", "level": "model", "family_id": "stability-ai-stable-diffusion", "developer": [ "stability-ai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": "Amazon Bedrock model ID stability.stable-image-ultra-v1:1, in-region in us-west-2 (Oregon) only.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "cross-region only", "status": "active", "summary": "Stability AI highest-fidelity text-to-image model for photorealistic, large-scale professional output.", "verified_on": "2026-10-12", "sources": [ "https://docs.aws.amazon.com/bedrock/latest/userguide/model-card-stability-ai-stable-image-ultra-1-0.html", "https://docs.aws.amazon.com/bedrock/latest/userguide/model-cards.html" ], "unverified": [] }, { "id": "stability-ai-stable-image-core", "name": "Stable Image Core", "level": "model", "family_id": "stability-ai-stable-diffusion", "developer": [ "stability-ai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": "Amazon Bedrock model ID stability.stable-image-core-v1:1, in-region in us-west-2 (Oregon) only.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "cross-region only", "status": "active", "summary": "Stability AI fast, affordable text-to-image model for rapid, high-volume content generation.", "verified_on": "2026-10-12", "sources": [ "https://docs.aws.amazon.com/bedrock/latest/userguide/model-card-stability-ai-stable-image-core-1-0.html", "https://docs.aws.amazon.com/bedrock/latest/userguide/model-cards.html" ], "unverified": [] }, { "id": "bria-ai-bria", "name": "Bria image models", "level": "family", "family_id": null, "developer": [ "bria-ai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Bria 2.3 is listed in Amazon Bedrock Marketplace and SageMaker JumpStart; Bria 3.2 and FIBO on Replicate.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Text-to-image models from Bria AI that the developer says are trained entirely on licensed data for commercial use.", "verified_on": "2026-10-12", "sources": [ "https://replicate.com/bria/image-3.2", "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-marketplace-model-reference.html", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json" ], "unverified": [] }, { "id": "bria-ai-fibo", "name": "Bria FIBO", "level": "model", "family_id": "bria-ai-bria", "developer": [ "bria-ai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": 8285836848, "params_active": null, "params_checkpoint": null, "params_note": "Gated download under the bria-fibo licence. Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2025-10-09", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open text-to-image model trained on licensed data that turns intent into structured JSON control for precise generation in enterprise and agent workflows.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/briaai/FIBO", "https://replicate.com/bria/fibo" ], "unverified": [] }, { "id": "amazon-nova-canvas", "name": "Amazon Nova Canvas", "level": "model", "family_id": "amazon-nova", "developer": [ "amazon" ], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "Model ID amazon.nova-canvas-v1:0; up to 4.19 million pixels; English prompts; regions US East (N. Virginia), Europe (Ireland), Asia Pacific (Tokyo).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "cross-region only", "status": "active", "summary": "Amazon image generation model with customisation and control that turns text and image inputs into images, offered through Amazon Bedrock.", "verified_on": "2026-10-12", "sources": [ "https://docs.aws.amazon.com/nova/latest/userguide/what-is-nova.html", "https://docs.aws.amazon.com/bedrock/latest/userguide/model-cards.html" ], "unverified": [] }, { "id": "amazon-nova-reel", "name": "Amazon Nova Reel", "level": "model", "family_id": "amazon-nova", "developer": [ "amazon" ], "model_class": "video generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "Model ID amazon.nova-reel-v1:1; 1280x720 at 24 fps; asynchronous API; regions US East (N. Virginia), Europe (Ireland), Asia Pacific (Tokyo).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "cross-region only", "status": "active", "summary": "Amazon video generation model that turns text and image inputs into short videos, offered through Amazon Bedrock.", "verified_on": "2026-10-12", "sources": [ "https://docs.aws.amazon.com/nova/latest/userguide/what-is-nova.html", "https://docs.aws.amazon.com/bedrock/latest/userguide/model-cards.html" ], "unverified": [] }, { "id": "google-deepmind-imagen-3", "name": "Imagen 3", "level": "model", "family_id": "google-deepmind-imagen", "developer": [ "google-deepmind" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-05-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "consumer app" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google latent diffusion text-to-image model, offered in ImageFX and on Vertex AI. Predecessor of Imagen 4.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2408.07009", "https://blog.google/technology/google-labs/video-image-generation-update-december-2024/", "https://replicate.com/google/imagen-3", "https://blog.google/technology/ai/google-generative-ai-veo-imagen-3/" ], "unverified": [] }, { "id": "google-deepmind-veo-2", "name": "Veo 2", "level": "model", "family_id": "google-deepmind-veo", "developer": [ "google-deepmind" ], "model_class": "video generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-12-16", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "consumer app" ], "deployment_tiers": [], "footprint_note": "Output up to 4K resolution per the announcement.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Google video generation model that follows simple and complex instructions and simulates real-world physics. Rolled out through VideoFX, YouTube and Vertex AI.", "verified_on": "2026-10-12", "sources": [ "https://blog.google/technology/google-labs/video-image-generation-update-december-2024/", "https://replicate.com/google/veo-2" ], "unverified": [] }, { "id": "openai-dall-e", "name": "DALL-E", "level": "family", "family_id": null, "developer": [ "openai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenAI text-to-image model line that preceded GPT Image.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2204.06125", "https://replicate.com/openai/dall-e-3" ], "unverified": [] }, { "id": "openai-dall-e-2", "name": "DALL-E 2", "level": "model", "family_id": "openai-dall-e", "developer": [ "openai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": "CLIP image-embedding prior plus diffusion decoder (unCLIP)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-04-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenAI text-to-image model that generates an image from a CLIP image embedding produced from the caption.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2204.06125", "https://replicate.com/openai/dall-e-2" ], "unverified": [] }, { "id": "bytedance-seedream-3-0", "name": "Seedream 3.0", "level": "model", "family_id": "bytedance-seedream", "developer": [ "bytedance" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-04-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": "Native 2K image generation (Replicate page).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ByteDance bilingual text-to-image model with native high-resolution output, the predecessor of Seedream 4.0.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2504.11346", "https://replicate.com/bytedance/seedream-3" ], "unverified": [] }, { "id": "bytedance-sdxl-lightning", "name": "SDXL-Lightning", "level": "model", "family_id": null, "developer": [ "bytedance" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": "SDXL distilled with progressive adversarial diffusion distillation", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-02-20", "release_date_basis": "registry date", "weights_date": "2024-02-20", "openness": "open weights restricted", "licence": "openrail-plus-plus", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Few-step text-to-image model distilled from SDXL that generates 1024px images in one to eight steps.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2402.13929", "https://huggingface.co/ByteDance/SDXL-Lightning", "https://replicate.com/bytedance/sdxl-lightning-4step" ], "unverified": [] }, { "id": "bytedance-pulid", "name": "PuLID", "level": "model", "family_id": null, "developer": [ "bytedance" ], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-04-24", "release_date_basis": "announcement page", "weights_date": "2024-05-01", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tuning-free identity customisation for text-to-image models that keeps a person's identity from a reference photo using contrastive alignment.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2404.16022", "https://huggingface.co/guozinan/PuLID", "https://replicate.com/bytedance/pulid" ], "unverified": [] }, { "id": "tencent-photomaker", "name": "PhotoMaker", "level": "model", "family_id": null, "developer": [ "tencent" ], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-12-07", "release_date_basis": "announcement page", "weights_date": "2024-01-13", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tencent ARC method that customises realistic human photos from reference images through a stacked identity embedding, without per-person fine-tuning.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2312.04461", "https://huggingface.co/TencentARC/PhotoMaker", "https://replicate.com/tencentarc/photomaker" ], "unverified": [] }, { "id": "ip-adapter", "name": "IP-Adapter (incl. FaceID)", "level": "model", "family_id": null, "developer": [ "tencent" ], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": "Decoupled cross-attention adapter for pretrained text-to-image diffusion models", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "IP-Adapter-FaceID checkpoints are a separate repository (h94/IP-Adapter-FaceID, created 2023-12-20).", "context_length": null, "release_date": "2023-08-13", "release_date_basis": "announcement page", "weights_date": "2023-08-16", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Lightweight adapter that lets a pretrained text-to-image diffusion model take an image as a prompt alongside text.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2308.06721", "https://huggingface.co/h94/IP-Adapter", "https://huggingface.co/api/models?pipeline_tag=text-to-image&sort=downloads&limit=50", "https://replicate.com/lucataco/ip-adapter-faceid", "https://arxiv.org/html/2308.06721" ], "unverified": [] }, { "id": "instantid", "name": "InstantID", "level": "model", "family_id": null, "developer": [], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-01-15", "release_date_basis": "announcement page", "weights_date": "2024-01-19", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Zero-shot identity-preserving image generation from a single face photo, released by the InstantX team.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2401.07519", "https://huggingface.co/InstantX/InstantID", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://replicate.com/zsxkib/instant-id" ], "unverified": [] }, { "id": "controlnet-family", "name": "ControlNet", "level": "family", "family_id": null, "developer": [ "stanford-university" ], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "ControlNet 1.0 and 1.1 condition Stable Diffusion 1.5 on canny edges, depth, pose, scribbles, segmentation and more; community versions exist for SDXL.", "context_length": null, "release_date": "2023-02-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Neural network structure that adds spatial conditioning such as edges, depth or pose to pretrained text-to-image diffusion models.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2302.05543", "https://github.com/open-mmlab/mmagic/tree/main/configs/controlnet", "https://aihub.qualcomm.com/models/controlnet_canny", "https://arxiv.org/html/2302.05543" ], "unverified": [] }, { "id": "controlnet-1-0", "name": "ControlNet 1.0", "level": "variant", "family_id": "controlnet-family", "developer": [ "stanford-university" ], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "OpenRAIL licence tag on Hugging Face. Qualcomm AI Hub ships a ControlNet (Canny) build for devices.", "context_length": null, "release_date": "2023-02-08", "release_date_basis": "registry date", "weights_date": "2023-02-08", "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download", "API" ], "deployment_tiers": [ "consumer GPU", "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First ControlNet release: conditioning models for Stable Diffusion 1.5 covering edges, depth, pose, scribbles and segmentation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2302.05543", "https://huggingface.co/lllyasviel/ControlNet", "https://aihub.qualcomm.com/models/controlnet_canny", "https://replicate.com/jagilley/controlnet-canny", "https://arxiv.org/html/2302.05543" ], "unverified": [] }, { "id": "controlnet-1-1", "name": "ControlNet 1.1", "level": "variant", "family_id": "controlnet-family", "developer": [ "stanford-university" ], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "OpenRAIL licence tag; 14 condition models such as control_v11p_sd15_openpose.", "context_length": null, "release_date": "2023-04-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Second ControlNet release with improved and additional conditioning models for Stable Diffusion 1.5.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/lllyasviel/ControlNet-v1-1", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://arxiv.org/html/2302.05543" ], "unverified": [] }, { "id": "xinsir-controlnet-union-sdxl", "name": "ControlNet Union SDXL 1.0", "level": "variant", "family_id": "controlnet-family", "developer": [], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": 1255958800, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2024-07-07", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Community ControlNet for SDXL that handles many control types in a single model.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/xinsir/controlnet-union-sdxl-1.0", "https://huggingface.co/api/models?pipeline_tag=text-to-image&sort=likes&limit=50" ], "unverified": [] }, { "id": "latent-consistency-model", "name": "Latent Consistency Models (LCM)", "level": "family", "family_id": null, "developer": [ "tsinghua-university" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-10-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Method that distils latent diffusion models into few-step generators that produce high-resolution images in about two to four steps.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2310.04378", "https://replicate.com/fofr/latent-consistency-model", "https://arxiv.org/html/2310.04378" ], "unverified": [] }, { "id": "lcm-dreamshaper-v7", "name": "LCM Dreamshaper v7", "level": "variant", "family_id": "latent-consistency-model", "developer": [ "tsinghua-university" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": 859602884, "params_active": null, "params_checkpoint": null, "params_note": "OpenVINO publishes int8, int4 and fp16 builds (OpenVINO/LCM_Dreamshaper_v7-int8-ov). Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2023-10-06", "release_date_basis": "announcement page", "weights_date": "2023-10-14", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU", "CPU server" ], "footprint_note": null, "runtimes": [ "openvino" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Latent consistency model distilled from Dreamshaper v7 for few-step text-to-image generation, also packaged by OpenVINO for CPUs.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2310.04378", "https://huggingface.co/SimianLuo/LCM_Dreamshaper_v7", "https://huggingface.co/OpenVINO/LCM_Dreamshaper_v7-int8-ov", "https://arxiv.org/html/2310.04378" ], "unverified": [] }, { "id": "playground-ai-playground-v2-5", "name": "Playground v2.5", "level": "model", "family_id": null, "developer": [ "playground-ai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": 2567463684, "params_active": null, "params_checkpoint": null, "params_note": "Playground v2.5 community licence. Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2024-02-16", "release_date_basis": "registry date", "weights_date": "2024-02-16", "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open text-to-image diffusion model tuned for aesthetic quality at 1024px.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2402.17245", "https://huggingface.co/playgroundai/playground-v2.5-1024px-aesthetic", "https://huggingface.co/api/models?pipeline_tag=text-to-image&sort=downloads&limit=50", "https://replicate.com/playgroundai/playground-v2.5-1024px-aesthetic" ], "unverified": [ "release_date" ] }, { "id": "nvidia-sana", "name": "SANA", "level": "family", "family_id": null, "developer": [ "nvidia" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-10-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA efficient text-to-image models using a linear diffusion transformer for images up to 4096x4096.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2410.10629", "https://replicate.com/nvidia/sana" ], "unverified": [] }, { "id": "nvidia-sana-1600m", "name": "SANA 1.6B", "level": "variant", "family_id": "nvidia-sana", "developer": [ "nvidia" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": "Linear diffusion transformer", "params_total": 1600000000, "params_active": null, "params_checkpoint": null, "params_note": "1.6B from the model name.", "context_length": null, "release_date": "2024-10-14", "release_date_basis": "announcement page", "weights_date": "2024-11-04", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Larger SANA text-to-image model at 1024px.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2410.10629", "https://huggingface.co/Efficient-Large-Model/Sana_1600M_1024px", "https://replicate.com/nvidia/sana" ], "unverified": [] }, { "id": "nvidia-sana-600m", "name": "SANA 0.6B", "level": "variant", "family_id": "nvidia-sana", "developer": [ "nvidia" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": 600000000, "params_active": null, "params_checkpoint": null, "params_note": "600M from the model name.", "context_length": null, "release_date": "2024-10-14", "release_date_basis": "announcement page", "weights_date": "2024-11-30", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Smaller SANA text-to-image model at 1024px.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2410.10629", "https://huggingface.co/Efficient-Large-Model/Sana_600M_1024px" ], "unverified": [] }, { "id": "nvidia-sana-sprint-1-6b", "name": "SANA-Sprint 1.6B", "level": "variant", "family_id": "nvidia-sana", "developer": [ "nvidia" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": 1600000000, "params_active": null, "params_checkpoint": null, "params_note": "1.6B from the model name.", "context_length": null, "release_date": "2025-03-12", "release_date_basis": "announcement page", "weights_date": "2025-03-21", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "One-step SANA variant trained with continuous-time consistency distillation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2503.09641", "https://huggingface.co/Efficient-Large-Model/Sana_Sprint_1.6B_1024px", "https://replicate.com/nvidia/sana-sprint-1.6b" ], "unverified": [] }, { "id": "sber-kandinsky-2-2", "name": "Kandinsky 2.2", "level": "model", "family_id": null, "developer": [ "sber" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": "Image prior plus latent diffusion decoder", "params_total": 1253057288, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2023-06-09", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open text-to-image model from Sber AI that maps text to an image embedding with a prior and decodes it with latent diffusion.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/kandinsky-community/kandinsky-2-2-decoder", "https://arxiv.org/abs/2310.03502", "https://replicate.com/ai-forever/kandinsky-2.2" ], "unverified": [ "release_date" ] }, { "id": "tencent-hunyuanimage-2-1", "name": "HunyuanImage 2.1", "level": "model", "family_id": "tencent-hunyuanimage", "developer": [ "tencent" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-09-05", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "tencent-hunyuan-community", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tencent open text-to-image model that generates 2K resolution images, predecessor of HunyuanImage 3.0.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/tencent/HunyuanImage-2.1", "https://replicate.com/tencent/hunyuan-image-2.1" ], "unverified": [] }, { "id": "tencent-hunyuan-dit", "name": "Hunyuan-DiT", "level": "model", "family_id": "tencent-hunyuanimage", "developer": [ "tencent" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": "Multi-resolution diffusion transformer with fine-grained Chinese and English understanding", "params_total": 1516534048, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2024-05-14", "release_date_basis": "announcement page", "weights_date": "2024-06-14", "openness": "open weights restricted", "licence": "tencent-hunyuan-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tencent bilingual Chinese and English text-to-image diffusion transformer.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2405.08748", "https://huggingface.co/Tencent-Hunyuan/HunyuanDiT-v1.1-Diffusers-Distilled", "https://huggingface.co/api/models?pipeline_tag=text-to-image&sort=downloads&limit=50" ], "unverified": [] }, { "id": "chroma1-hd", "name": "Chroma1-HD", "level": "model", "family_id": null, "developer": [], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": 8899983424, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2025-08-08", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Community open text-to-image model released by lodestones on Hugging Face and listed in the Azure AI Foundry catalogue.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/lodestones/Chroma1-HD", "https://ai.azure.com/catalog/models?task=text-to-image" ], "unverified": [] }, { "id": "alibaba-ovis-image-7b", "name": "Ovis-Image 7B", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": 7370449728, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2025-11-28", "release_date_basis": "announcement page", "weights_date": "2025-11-28", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open 7B text-to-image model built on Ovis2.5 and listed in the Azure AI Foundry catalogue.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2511.22982", "https://huggingface.co/ATH-MaaS/Ovis-Image-7B", "https://ai.azure.com/catalog/models?task=text-to-image" ], "unverified": [] }, { "id": "deci-decidiffusion-1-0", "name": "DeciDiffusion 1.0", "level": "model", "family_id": null, "developer": [ "deci" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": 859520964, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2023-09-13", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "openrail-plus-plus", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Deci text-to-image latent diffusion model designed for faster inference, listed in the Azure AI Foundry catalogue.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Deci/DeciDiffusion-v1-0", "https://ai.azure.com/catalog/models?task=text-to-image" ], "unverified": [] }, { "id": "prompthero-openjourney", "name": "Openjourney", "level": "model", "family_id": null, "developer": [], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": 123060557, "params_active": null, "params_checkpoint": null, "params_note": "CreativeML OpenRAIL-M licence tag. Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2022-11-08", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Community Stable Diffusion fine-tune from PromptHero, listed in the Azure AI Foundry catalogue.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/prompthero/openjourney", "https://ai.azure.com/catalog/models?task=text-to-image" ], "unverified": [] }, { "id": "rundiffusion-juggernaut-xl-v9", "name": "Juggernaut-XL v9", "level": "model", "family_id": null, "developer": [], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "CreativeML OpenRAIL-M licence tag. OpenVINO publishes int4, int8 and fp16 builds.", "context_length": null, "release_date": "2024-02-18", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [ "openvino" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Community fine-tune of SDXL from RunDiffusion, also packaged by OpenVINO for Intel hardware.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/RunDiffusion/Juggernaut-XL-v9", "https://huggingface.co/OpenVINO/Juggernaut-XL-v9-fp16-ov" ], "unverified": [] }, { "id": "instruct-pix2pix", "name": "InstructPix2Pix", "level": "model", "family_id": null, "developer": [ "uc-berkeley" ], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": 859532484, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2022-11-17", "release_date_basis": "announcement page", "weights_date": "2023-01-20", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Diffusion model that edits an image by following a written instruction such as change the sky to sunset.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2211.09800", "https://huggingface.co/timbrooks/instruct-pix2pix", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://www.kaggle.com/models/timothy-brooks/instruct-pix2pix", "https://arxiv.org/html/2211.09800" ], "unverified": [] }, { "id": "idm-vton", "name": "IDM-VTON", "level": "model", "family_id": null, "developer": [ "kaist", "omnious" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-03-08", "release_date_basis": "announcement page", "weights_date": "2024-03-28", "openness": "open weights restricted", "licence": "cc-by-nc-sa-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Diffusion model for virtual try-on that dresses a person photo in a given garment image, aimed at fashion retail.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2403.05139", "https://huggingface.co/yisol/IDM-VTON", "https://huggingface.co/api/models?pipeline_tag=image-to-image&sort=likes&limit=50", "https://arxiv.org/html/2403.05139" ], "unverified": [] }, { "id": "nvidia-chronoedit-14b", "name": "ChronoEdit-14B", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": 16394878784, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2025-10-05", "release_date_basis": "announcement page", "weights_date": "2025-10-28", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA image editing model that treats editing as temporal reasoning to keep edits physically consistent.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2510.04290", "https://huggingface.co/nvidia/ChronoEdit-14B-Diffusers", "https://huggingface.co/api/models?pipeline_tag=other&sort=likes&limit=50" ], "unverified": [] }, { "id": "jd-com-joyai-image-edit", "name": "JoyAI-Image-Edit", "level": "model", "family_id": null, "developer": [ "jd-com" ], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": 16263675968, "params_active": null, "params_checkpoint": null, "params_note": "JD also released JoyAI-Video-Edit. Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2026-04-10", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "JD.com open instruction-based image editing model.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/jdopensource/JoyAI-Image-Edit-Diffusers", "https://huggingface.co/api/models?pipeline_tag=image-to-image&sort=downloads&limit=50" ], "unverified": [] }, { "id": "stepfun-step1x-edit", "name": "Step1X-Edit", "level": "model", "family_id": null, "developer": [ "stepfun" ], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-04-23", "release_date_basis": "registry date", "weights_date": "2025-04-23", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "StepFun open framework for general image editing from text instructions.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2504.17761", "https://huggingface.co/stepfun-ai/Step1X-Edit", "https://replicate.com/zsxkib/step1x-edit" ], "unverified": [] }, { "id": "ic-light", "name": "IC-Light", "level": "model", "family_id": null, "developer": [], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-05-08", "release_date_basis": "registry date", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Image relighting model that changes the lighting of a foreground subject, used for product photos.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/lllyasviel/ic-light", "https://www.modelscope.cn/models/AI-ModelScope/ic-light" ], "unverified": [ "summary" ] }, { "id": "apple-starflow", "name": "STARFlow", "level": "model", "family_id": null, "developer": [ "apple" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "other", "architecture_detail": "Latent normalising flow", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Apple AMLR licence tag. The repository also covers STARFlow-V for video (arXiv 2511.20462).", "context_length": null, "release_date": "2025-06-06", "release_date_basis": "announcement page", "weights_date": "2025-11-26", "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Apple text-to-image model that scales latent normalising flows to high-resolution image synthesis.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2506.06276", "https://huggingface.co/apple/starflow" ], "unverified": [] }, { "id": "nvidia-maisi", "name": "MAISI (Medical AI for Synthetic Imaging)", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "NGC MONAI toolkit entries also include BraTS MRI generative diffusion models.", "context_length": null, "release_date": "2024-09-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Diffusion model that generates synthetic 3D CT volumes for medical imaging research and data augmentation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2409.11169", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/monaitoolkit/models/monai_brats_mri_axial_slices_generative_diffusion" ], "unverified": [] }, { "id": "dcgan", "name": "DCGAN", "level": "model", "family_id": null, "developer": [ "indico", "meta" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2015-11-19", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Deep convolutional generative adversarial network, a foundational architecture for GAN image synthesis.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1511.06434", "https://github.com/open-mmlab/mmagic/tree/main/configs/dcgan", "https://pytorch.org/hub/facebookresearch_pytorch-gan-zoo_dcgan", "https://arxiv.org/html/1511.06434" ], "unverified": [] }, { "id": "pggan", "name": "Progressive GAN (PGGAN)", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2017-10-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "GAN trained by progressively growing generator and discriminator resolution, used for high-resolution face synthesis.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1710.10196", "https://github.com/open-mmlab/mmagic/tree/main/configs/pggan", "https://www.kaggle.com/models/google/progan-128", "https://arxiv.org/html/1710.10196" ], "unverified": [] }, { "id": "nvidia-stylegan-family", "name": "StyleGAN", "level": "family", "family_id": null, "developer": [ "nvidia" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA style-based GAN line used for face and image synthesis and synthetic data.", "verified_on": "2026-10-12", "sources": [ "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/research/models/stylegan2", "https://github.com/NVlabs/stylegan" ], "unverified": [] }, { "id": "nvidia-stylegan", "name": "StyleGAN", "level": "model", "family_id": "nvidia-stylegan-family", "developer": [ "nvidia" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2018-12-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Style-based generator architecture for GANs that separates high-level attributes from stochastic detail.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1812.04948", "https://github.com/open-mmlab/mmagic/tree/main/configs/styleganv1", "https://github.com/NVlabs/stylegan" ], "unverified": [] }, { "id": "nvidia-stylegan2", "name": "StyleGAN2", "level": "model", "family_id": "nvidia-stylegan-family", "developer": [ "nvidia" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2019-12-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Revised StyleGAN that removes characteristic artefacts and improves image quality.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1912.04958", "https://github.com/open-mmlab/mmagic/tree/main/configs/styleganv2", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/research/models/stylegan2" ], "unverified": [] }, { "id": "nvidia-stylegan3", "name": "StyleGAN3", "level": "model", "family_id": "nvidia-stylegan-family", "developer": [ "nvidia" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2021-06-23", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alias-free GAN that makes generated details move consistently with the depicted surfaces.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2106.12423", "https://github.com/open-mmlab/mmagic/tree/main/configs/styleganv3", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/research/models/stylegan2" ], "unverified": [] }, { "id": "google-deepmind-biggan", "name": "BigGAN (incl. BigGAN-Deep)", "level": "model", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "TF Hub and Kaggle publish BigGAN at several resolutions under Apache 2.0. Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2018-09-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Large-scale class-conditional GAN trained on ImageNet for high-fidelity image synthesis.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1809.11096", "https://www.kaggle.com/models/deepmind/biggan", "https://github.com/open-mmlab/mmagic/tree/main/configs/biggan" ], "unverified": [] }, { "id": "google-deepmind-bigbigan", "name": "BigBiGAN", "level": "model", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Published on Kaggle Models (TF Hub) under Apache 2.0. ", "context_length": null, "release_date": "2019-07-04", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "BigGAN extended with an encoder for unsupervised image representation learning.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1907.02544", "https://www.kaggle.com/models/deepmind/bigbigan" ], "unverified": [] }, { "id": "sagan", "name": "SAGAN (Self-Attention GAN)", "level": "model", "family_id": null, "developer": [ "rutgers-university", "google" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2018-05-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "GAN that adds self-attention so the generator can model long-range dependencies.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1805.08318", "https://github.com/open-mmlab/mmagic/tree/main/configs/sagan", "https://arxiv.org/html/1805.08318" ], "unverified": [] }, { "id": "singan", "name": "SinGAN", "level": "model", "family_id": null, "developer": [ "technion", "google" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2019-05-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "GAN trained on a single natural image that generates varied samples with the same patch statistics.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1905.01164", "https://github.com/open-mmlab/mmagic/tree/main/configs/singan", "https://arxiv.org/html/1905.01164" ], "unverified": [] }, { "id": "vq-vae", "name": "VQ-VAE", "level": "model", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-11-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Variational autoencoder with discrete latent codes, used as an image and audio tokeniser in later generative models.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1711.00937", "https://paperswithcode.com/method/vq-vae", "https://arxiv.org/html/1711.00937" ], "unverified": [] }, { "id": "pix2pix", "name": "pix2pix", "level": "model", "family_id": null, "developer": [ "uc-berkeley" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2016-11-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Conditional GAN for paired image-to-image translation such as maps to aerial photos or sketches to images.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1611.07004", "https://github.com/open-mmlab/mmagic/tree/main/configs/pix2pix", "https://modelzoo.co/model/pytorch-cyclegan-and-pix2pix", "https://arxiv.org/html/1611.07004" ], "unverified": [] }, { "id": "cyclegan", "name": "CycleGAN", "level": "model", "family_id": null, "developer": [ "uc-berkeley" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Also in the ONNX Model Zoo. Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2017-03-30", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Unpaired image-to-image translation with cycle-consistent GANs, used for style and domain transfer.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1703.10593", "https://github.com/open-mmlab/mmagic/tree/main/configs/cyclegan", "https://modelzoo.co/model/pytorch-cyclegan-and-pix2pix", "https://github.com/onnx/models/blob/main/README.md", "https://arxiv.org/html/1703.10593" ], "unverified": [] }, { "id": "stargan", "name": "StarGAN (v1, v2)", "level": "model", "family_id": null, "developer": [ "korea-university", "naver" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "StarGAN v2 (arXiv 1912.01865) adds diverse styles per domain. ", "context_length": null, "release_date": "2017-11-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Single GAN that translates images between multiple domains, for example facial attributes.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1711.09020", "https://arxiv.org/abs/1912.01865", "https://modelzoo.co/model/stargan", "https://arxiv.org/html/1711.09020" ], "unverified": [] }, { "id": "nvidia-pix2pixhd", "name": "pix2pixHD", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2017-11-30", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA conditional GAN that synthesises high-resolution photorealistic images from semantic label maps.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1711.11585", "https://catalog.ngc.nvidia.com/orgs/nvidia/models/pix2pixhd", "https://modelzoo.co/model/vid2vid" ], "unverified": [] }, { "id": "nvidia-vid2vid", "name": "vid2vid", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "video generation", "modalities_in": [ "video" ], "modalities_out": [ "video" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-08-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA video-to-video synthesis that turns segmentation or pose sequences into photorealistic video.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1808.06601", "https://github.com/NVIDIA/vid2vid", "https://modelzoo.co/model/vid2vid" ], "unverified": [] }, { "id": "cocosnet", "name": "CoCosNet", "level": "model", "family_id": null, "developer": [ "university-of-science-and-technology-of-china", "microsoft-research" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "In the OpenVINO Open Model Zoo (public/cocosnet). ", "context_length": null, "release_date": "2020-04-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [ "openvino" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Exemplar-based image translation that learns dense correspondence between domains to transfer an exemplar style.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2004.05571", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/cocosnet", "https://arxiv.org/html/2004.05571" ], "unverified": [] }, { "id": "colorization-zhang", "name": "Colorful Image Colorization", "level": "model", "family_id": null, "developer": [ "uc-berkeley" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "OpenVINO Open Model Zoo ships colorization-siggraph and colorization-v2. ", "context_length": null, "release_date": "2016-03-28", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [ "openvino" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "CNN that predicts plausible colours for grayscale photos.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1603.08511", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/colorization-siggraph", "https://arxiv.org/html/1603.08511" ], "unverified": [] }, { "id": "instcolorization", "name": "InstColorization", "level": "model", "family_id": null, "developer": [ "national-tsing-hua-university" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2020-05-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Instance-aware image colourisation that colours detected objects separately before fusing them.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2005.10825", "https://github.com/open-mmlab/mmagic/tree/main/configs/inst_colorization", "https://arxiv.org/html/2005.10825" ], "unverified": [] }, { "id": "ddcolor", "name": "DDColor", "level": "model", "family_id": null, "developer": [ "alibaba-damo" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "other", "architecture_detail": "Dual decoders (pixel and query-based colour)", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Qualcomm AI Hub ships a device build. ", "context_length": null, "release_date": "2022-12-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Photo-realistic image colourisation with dual decoders.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2212.11613", "https://aihub.qualcomm.com/models/ddcolor", "https://arxiv.org/html/2212.11613" ], "unverified": [] }, { "id": "fast-neural-style", "name": "Fast Neural Style Transfer", "level": "model", "family_id": null, "developer": [ "stanford-university" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "ONNX Model Zoo and OpenVINO (fast-neural-style-mosaic-onnx) builds. ", "context_length": null, "release_date": "2016-03-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [ "onnx-runtime", "openvino" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Feed-forward network trained with perceptual losses for real-time style transfer.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1603.08155", "https://github.com/onnx/models/tree/main/validated/vision/style_transfer/fast_neural_style", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/fast-neural-style-mosaic-onnx", "https://arxiv.org/html/1603.08155" ], "unverified": [] }, { "id": "google-arbitrary-image-stylization", "name": "Arbitrary Image Stylization (Magenta)", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Kaggle Models (TF Hub) arbitrary-image-stylization-v1, Apache 2.0. ", "context_length": null, "release_date": "2017-05-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time style transfer network that applies the style of any painting to a photo.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1705.06830", "https://www.kaggle.com/models/google/arbitrary-image-stylization-v1" ], "unverified": [] }, { "id": "google-boundless", "name": "Boundless", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Kaggle Models (TF Hub), Apache 2.0. ", "context_length": null, "release_date": "2019-08-19", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "GAN for image extension (outpainting) beyond the original borders.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1908.07007", "https://www.kaggle.com/models/google/compare-gan" ], "unverified": [] }, { "id": "google-deepmind-spiral", "name": "SPIRAL", "level": "model", "family_id": null, "developer": [ "google-deepmind" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "other", "architecture_detail": "Reinforcement learning agent trained adversarially to drive a painting program", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2018-04-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Agent that synthesises images by writing programs for a painting environment, trained with reinforced adversarial learning.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1804.01118", "https://www.kaggle.com/models/deepmind/spiral" ], "unverified": [] }, { "id": "cp-vton", "name": "CP-VTON", "level": "model", "family_id": null, "developer": [ "sun-yat-sen-university", "sensetime" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "In OpenMMLab MMFashion (Apache-2.0). ", "context_length": null, "release_date": "2018-07-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Image-based virtual try-on network that preserves garment characteristics when fitting clothes to a person.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1807.07688", "https://github.com/open-mmlab/mmfashion/tree/master/docs/MODEL_ZOO.md", "https://arxiv.org/html/1807.07688" ], "unverified": [] }, { "id": "spgan", "name": "SPGAN", "level": "model", "family_id": null, "developer": [ "university-of-chinese-academy-of-sciences", "university-of-technology-sydney" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "In OpenMMLab OpenUnReID (Apache-2.0). ", "context_length": null, "release_date": "2017-11-19", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Image-to-image domain adaptation GAN that translates labelled images to a new camera domain for person re-identification.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1711.07027", "https://github.com/open-mmlab/OpenUnReID/tree/master/tools/SPGAN", "https://arxiv.org/html/1711.07027" ], "unverified": [] }, { "id": "deepfill", "name": "DeepFill", "level": "family", "family_id": null, "developer": [ "uiuc", "adobe" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Generative image inpainting models with contextual attention (v1) and gated convolution (v2).", "verified_on": "2026-10-12", "sources": [ "https://github.com/open-mmlab/mmagic/tree/main/configs/deepfillv1", "https://github.com/open-mmlab/mmagic/tree/main/configs/deepfillv2", "https://arxiv.org/html/1801.07892" ], "unverified": [] }, { "id": "deepfill-v1", "name": "DeepFill v1", "level": "variant", "family_id": "deepfill", "developer": [ "uiuc", "adobe" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2018-01-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Generative inpainting with contextual attention that copies features from distant image regions.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1801.07892", "https://github.com/open-mmlab/mmagic/tree/main/configs/deepfillv1", "https://arxiv.org/html/1801.07892" ], "unverified": [] }, { "id": "deepfill-v2", "name": "DeepFill v2", "level": "variant", "family_id": "deepfill", "developer": [ "uiuc", "adobe" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2018-06-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Free-form inpainting with gated convolution, supporting user-guided masks and sketches.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1806.03589", "https://github.com/open-mmlab/mmagic/tree/main/configs/deepfillv2", "https://arxiv.org/html/1806.03589" ], "unverified": [] }, { "id": "partial-convolution-inpainting", "name": "Partial Convolution Inpainting", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2018-04-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Inpainting with partial convolutions that handles irregular holes.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1804.07723", "https://github.com/open-mmlab/mmagic/tree/main/configs/partial_conv", "https://arxiv.org/html/1804.07723" ], "unverified": [] }, { "id": "gmcnn", "name": "GMCNN", "level": "model", "family_id": null, "developer": [ "cuhk", "tencent" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "OpenVINO Open Model Zoo gmcnn-places2-tf. ", "context_length": null, "release_date": "2018-10-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [ "openvino" ], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Generative multi-column CNN for image inpainting.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1810.08771", "https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/gmcnn-places2-tf", "https://arxiv.org/html/1810.08771" ], "unverified": [] }, { "id": "aot-gan", "name": "AOT-GAN", "level": "model", "family_id": null, "developer": [ "sun-yat-sen-university", "microsoft-research" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Qualcomm AI Hub ships a device build. Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2021-04-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "High-resolution inpainting with aggregated contextual transformations.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2104.01431", "https://github.com/open-mmlab/mmagic/tree/main/configs/aot_gan", "https://aihub.qualcomm.com/models/aotgan", "https://arxiv.org/html/2104.01431" ], "unverified": [] }, { "id": "lama-inpainting", "name": "LaMa", "level": "model", "family_id": null, "developer": [ "samsung", "skoltech" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": "Fast Fourier convolutions", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Qualcomm AI Hub (lama_dilated), ModelScope (Apache 2.0 port) and Vertex AI Model Garden builds. ", "context_length": null, "release_date": "2021-09-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Large-mask inpainting with Fourier convolutions, widely used for object removal.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2109.07161", "https://aihub.qualcomm.com/models/lama_dilated", "https://www.modelscope.cn/models/iic/cv_fft_inpainting_lama", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden", "https://arxiv.org/html/2109.07161" ], "unverified": [] }, { "id": "powerpaint", "name": "PowerPaint", "level": "model", "family_id": null, "developer": [ "tsinghua-university", "shanghai-ai-lab" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2023-12-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Text-guided inpainting and outpainting model that uses task prompts for object insertion, removal and shape-guided filling.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2312.03594", "https://github.com/open-mmlab/mmagic/tree/main/projects/powerpaint", "https://arxiv.org/html/2312.03594" ], "unverified": [] }, { "id": "srcnn", "name": "SRCNN", "level": "model", "family_id": null, "developer": [ "cuhk" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2014-12-31", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First deep convolutional network for single-image super-resolution.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1501.00092", "https://github.com/open-mmlab/mmagic/tree/main/configs/srcnn", "https://arxiv.org/html/1501.00092" ], "unverified": [] }, { "id": "srgan", "name": "SRGAN and SRResNet", "level": "model", "family_id": null, "developer": [ "twitter" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Qualcomm AI Hub ships an SRGAN build. Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2016-09-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "GAN-based photo-realistic single-image super-resolution and its residual network generator.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1609.04802", "https://github.com/open-mmlab/mmagic/tree/main/configs/srgan_resnet", "https://aihub.qualcomm.com/models/srgan", "https://arxiv.org/html/1609.04802" ], "unverified": [] }, { "id": "edsr", "name": "EDSR", "level": "model", "family_id": null, "developer": [ "seoul-national-university" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2017-07-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Enhanced deep residual network for single-image super-resolution.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1707.02921", "https://github.com/open-mmlab/mmagic/tree/main/configs/edsr", "https://arxiv.org/html/1707.02921" ], "unverified": [] }, { "id": "rdn", "name": "RDN", "level": "model", "family_id": null, "developer": [ "northeastern-university" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2018-02-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Residual dense network for image super-resolution.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1802.08797", "https://github.com/open-mmlab/mmagic/tree/main/configs/rdn", "https://arxiv.org/html/1802.08797" ], "unverified": [] }, { "id": "esrgan-family", "name": "ESRGAN", "level": "family", "family_id": null, "developer": [ "cuhk", "ntu-singapore" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "GAN super-resolution line from ESRGAN to Real-ESRGAN, the most used practical image upscalers.", "verified_on": "2026-10-12", "sources": [ "https://github.com/open-mmlab/mmagic/tree/main/configs/esrgan", "https://github.com/open-mmlab/mmagic/tree/main/configs/real_esrgan", "https://arxiv.org/html/1809.00219" ], "unverified": [] }, { "id": "esrgan", "name": "ESRGAN", "level": "variant", "family_id": "esrgan-family", "developer": [ "cuhk", "ntu-singapore" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Qualcomm AI Hub ships a build. Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2018-09-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Enhanced super-resolution GAN with residual-in-residual dense blocks.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1809.00219", "https://github.com/open-mmlab/mmagic/tree/main/configs/esrgan", "https://aihub.qualcomm.com/models/esrgan", "https://arxiv.org/html/1809.00219" ], "unverified": [] }, { "id": "real-esrgan", "name": "Real-ESRGAN", "level": "variant", "family_id": "esrgan-family", "developer": [ "tencent" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Qualcomm AI Hub ships x4plus and general x4v3 builds. Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2021-07-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Blind super-resolution model trained on synthetic degradations for real-world photos.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2107.10833", "https://github.com/open-mmlab/mmagic/tree/main/configs/real_esrgan", "https://aihub.qualcomm.com/models/real_esrgan_x4plus", "https://arxiv.org/html/2107.10833" ], "unverified": [] }, { "id": "swinir", "name": "SwinIR", "level": "model", "family_id": null, "developer": [ "eth-zurich" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": "Swin Transformer", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2021-08-23", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Image restoration model based on the Swin Transformer for super-resolution, denoising and JPEG artefact removal.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2108.10257", "https://github.com/open-mmlab/mmagic/tree/main/configs/swinir", "https://arxiv.org/html/2108.10257" ], "unverified": [] }, { "id": "nafnet", "name": "NAFNet", "level": "model", "family_id": null, "developer": [ "megvii" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2022-04-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Simple nonlinear-activation-free baseline network for image restoration such as deblurring and denoising.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2204.04676", "https://github.com/open-mmlab/mmagic/tree/main/configs/nafnet", "https://arxiv.org/html/2204.04676" ], "unverified": [] }, { "id": "restormer", "name": "Restormer", "level": "model", "family_id": null, "developer": [ "inception-institute-of-ai", "mbzuai" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2021-11-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Efficient transformer for high-resolution image restoration.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2111.09881", "https://github.com/open-mmlab/mmagic/tree/main/configs/restormer", "https://arxiv.org/html/2111.09881" ], "unverified": [] }, { "id": "glean", "name": "GLEAN", "level": "model", "family_id": null, "developer": [ "ntu-singapore" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2020-12-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Large-factor image super-resolution that uses a pretrained GAN as a latent bank.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2012.00739", "https://github.com/open-mmlab/mmagic/tree/main/configs/glean", "https://arxiv.org/html/2012.00739" ], "unverified": [] }, { "id": "liif", "name": "LIIF", "level": "model", "family_id": null, "developer": [ "uc-san-diego", "nvidia" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "other", "architecture_detail": "Local implicit image function", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2020-12-16", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Continuous image representation that allows super-resolution at arbitrary scales.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2012.09161", "https://github.com/open-mmlab/mmagic/tree/main/configs/liif", "https://arxiv.org/html/2012.09161" ], "unverified": [] }, { "id": "ttsr", "name": "TTSR", "level": "model", "family_id": null, "developer": [ "shanghai-jiao-tong-university", "microsoft-research" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2020-06-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Texture transformer network for reference-based image super-resolution.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2006.04139", "https://github.com/open-mmlab/mmagic/tree/main/configs/ttsr", "https://arxiv.org/html/2006.04139" ], "unverified": [] }, { "id": "dic-face-sr", "name": "DIC (Deep Iterative Collaboration)", "level": "model", "family_id": null, "developer": [ "tsinghua-university" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2020-03-29", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Face super-resolution that iterates between image recovery and facial landmark estimation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2003.13063", "https://github.com/open-mmlab/mmagic/tree/main/configs/dic", "https://arxiv.org/html/2003.13063" ], "unverified": [] }, { "id": "vico", "name": "ViCo", "level": "model", "family_id": null, "developer": [ "hku" ], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2023-06-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Plug-and-play visual condition for personalised text-to-image generation from a few subject images.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2306.00971", "https://github.com/open-mmlab/mmagic/tree/main/configs/vico", "https://arxiv.org/html/2306.00971" ], "unverified": [] }, { "id": "draggan", "name": "DragGAN", "level": "model", "family_id": null, "developer": [ "max-planck-institute-for-informatics" ], "model_class": "image generation", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2023-05-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Interactive point-based image manipulation by dragging handle points on the GAN image manifold.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2305.10973", "https://github.com/open-mmlab/mmagic/tree/main/configs/draggan", "https://arxiv.org/html/2305.10973" ], "unverified": [] }, { "id": "fastcomposer", "name": "FastComposer", "level": "model", "family_id": null, "developer": [ "massachusetts-institute-of-technology" ], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2023-05-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tuning-free multi-subject image generation with localised attention.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2305.10431", "https://github.com/open-mmlab/mmagic/tree/main/configs/fastcomposer", "https://arxiv.org/html/2305.10431" ], "unverified": [] }, { "id": "styleshot", "name": "StyleShot", "level": "model", "family_id": null, "developer": [ "tongji-university", "shanghai-ai-lab" ], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-07-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Style transfer to any content from a reference style image without fine-tuning.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2407.01414", "https://github.com/open-mmlab/StyleShot", "https://arxiv.org/html/2407.01414" ], "unverified": [] }, { "id": "anycontrol", "name": "AnyControl", "level": "model", "family_id": null, "developer": [ "shanghai-ai-lab" ], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-06-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Text-to-image generation with versatile, combined spatial controls.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2406.18958", "https://github.com/open-mmlab/AnyControl", "https://arxiv.org/html/2406.18958" ], "unverified": [] }, { "id": "ddpm", "name": "DDPM (Denoising Diffusion Probabilistic Models)", "level": "model", "family_id": null, "developer": [ "uc-berkeley" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": 35746307, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2020-06-19", "release_date_basis": "announcement page", "weights_date": "2022-06-16", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Foundational denoising diffusion model for image generation; checkpoints such as ddpm-cifar10-32 are on Hugging Face.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2006.11239", "https://huggingface.co/google/ddpm-cifar10-32", "https://huggingface.co/api/models?pipeline_tag=unconditional-image-generation&sort=downloads&limit=50", "https://arxiv.org/html/2006.11239" ], "unverified": [] }, { "id": "improved-ddpm", "name": "Improved DDPM", "level": "model", "family_id": null, "developer": [ "openai" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2021-02-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Improved denoising diffusion model with learned variances and better sampling efficiency.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2102.09672", "https://github.com/open-mmlab/mmagic/tree/main/https://github.com/open-mmlab/mmgeneration/tree/master/configs/improved_ddpm", "https://arxiv.org/html/2102.09672" ], "unverified": [] }, { "id": "guided-diffusion-adm", "name": "Guided Diffusion (ADM)", "level": "model", "family_id": null, "developer": [ "openai" ], "model_class": "image generation", "modalities_in": [], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2021-05-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Diffusion model with classifier guidance that surpassed GANs on ImageNet synthesis.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2105.05233", "https://github.com/open-mmlab/mmagic/tree/main/configs/guided_diffusion", "https://arxiv.org/html/2105.05233" ], "unverified": [] }, { "id": "glide", "name": "GLIDE", "level": "model", "family_id": null, "developer": [ "openai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2021-12-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Text-guided diffusion model for photorealistic image generation and editing; a small filtered model was released.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2112.10741", "https://github.com/open-mmlab/mmagic/tree/main/projects/glide/configs", "https://arxiv.org/html/2112.10741" ], "unverified": [] }, { "id": "latent-diffusion-ldm", "name": "Latent Diffusion (LDM)", "level": "model", "family_id": null, "developer": [ "lmu-munich", "runway" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-12-20", "release_date_basis": "announcement page", "weights_date": "2022-07-18", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Diffusion in the latent space of a pretrained autoencoder, the method behind Stable Diffusion.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2112.10752", "https://huggingface.co/CompVis/ldm-text2im-large-256", "https://huggingface.co/api/models?pipeline_tag=unconditional-image-generation&sort=downloads&limit=50", "https://arxiv.org/html/2112.10752" ], "unverified": [] }, { "id": "stability-ai-stable-video-diffusion", "name": "Stable Video Diffusion", "level": "family", "family_id": null, "developer": [ "stability-ai" ], "model_class": "video generation", "modalities_in": [ "image" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-11-25", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Stability AI latent video diffusion models that turn a single image into a short video clip.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2311.15127", "https://huggingface.co/api/models?pipeline_tag=image-to-video&sort=downloads&limit=50", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden" ], "unverified": [] }, { "id": "stability-ai-svd-img2vid", "name": "SVD img2vid", "level": "variant", "family_id": "stability-ai-stable-video-diffusion", "developer": [ "stability-ai" ], "model_class": "video generation", "modalities_in": [ "image" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": null, "params_total": 1524623082, "params_active": null, "params_checkpoint": null, "params_note": "Stable Video Diffusion community licence. Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2023-11-20", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "single datacentre GPU", "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "First Stable Video Diffusion image-to-video checkpoint.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/stabilityai/stable-video-diffusion-img2vid", "https://arxiv.org/abs/2311.15127" ], "unverified": [] }, { "id": "stability-ai-svd-img2vid-xt", "name": "SVD img2vid-XT", "level": "variant", "family_id": "stability-ai-stable-video-diffusion", "developer": [ "stability-ai" ], "model_class": "video generation", "modalities_in": [ "image" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": null, "params_total": 1524623082, "params_active": null, "params_checkpoint": null, "params_note": "Stable Video Diffusion community licence. Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2023-11-20", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "single datacentre GPU", "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Stable Video Diffusion checkpoint fine-tuned to generate longer clips (25 frames).", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/stabilityai/stable-video-diffusion-img2vid-xt", "https://arxiv.org/abs/2311.15127" ], "unverified": [] }, { "id": "stability-ai-svd-img2vid-xt-1-1", "name": "SVD img2vid-XT 1.1", "level": "variant", "family_id": "stability-ai-stable-video-diffusion", "developer": [ "stability-ai" ], "model_class": "video generation", "modalities_in": [ "image" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": null, "params_total": 1524623082, "params_active": null, "params_checkpoint": null, "params_note": "Stable Video Diffusion community licence. Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2024-02-02", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "single datacentre GPU", "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Updated XT checkpoint released in 2024.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/stabilityai/stable-video-diffusion-img2vid-xt-1-1", "https://arxiv.org/abs/2311.15127" ], "unverified": [] }, { "id": "alibaba-wan-2-1-t2v-1-3b", "name": "Wan 2.1 T2V 1.3B", "level": "variant", "family_id": "alibaba-wan", "developer": [ "alibaba" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": null, "params_total": 1418996800, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2025-02-25", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [ "consumer GPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Smallest Wan 2.1 text-to-video model, offered in Amazon SageMaker JumpStart.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Wan-AI/Wan2.1-T2V-1.3B-Diffusers", "https://arxiv.org/abs/2503.20314", "https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json", "https://github.com/Wan-Video/Wan2.1" ], "unverified": [ "summary" ] }, { "id": "alibaba-wan-2-1-vace", "name": "Wan2.1-VACE (1.3B, 14B)", "level": "model", "family_id": "alibaba-wan", "developer": [ "alibaba" ], "model_class": "video generation", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": null, "params_total": 17337592896, "params_active": null, "params_checkpoint": null, "params_note": "Also a 1.3B size (Wan-AI/Wan2.1-VACE-1.3B). Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2025-03-10", "release_date_basis": "announcement page", "weights_date": "2025-05-13", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "All-in-one Wan 2.1 model for video creation and editing tasks such as reference-to-video, video-to-video and masked editing.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2503.07598", "https://huggingface.co/Wan-AI/Wan2.1-VACE-14B", "https://www.modelscope.cn/models/Wan-AI/Wan2.1-VACE-14B" ], "unverified": [] }, { "id": "alibaba-wan-2-2-s2v-14b", "name": "Wan2.2-S2V-14B", "level": "model", "family_id": "alibaba-wan-2-2", "developer": [ "alibaba" ], "model_class": "video generation", "modalities_in": [ "image", "audio", "text" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": null, "params_total": 16295755609, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2025-08-25", "release_date_basis": "registry date", "weights_date": "2025-08-25", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Audio-driven Wan 2.2 model that generates cinematic video of a character from an image and a speech or audio track.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2508.18621", "https://huggingface.co/Wan-AI/Wan2.2-S2V-14B", "https://www.modelscope.cn/models/Wan-AI/Wan2.2-S2V-14B" ], "unverified": [] }, { "id": "alibaba-wan-fun", "name": "Wan-Fun (Wan2.1-Fun, Wan2.2-Fun)", "level": "model", "family_id": "alibaba-wan", "developer": [ "alibaba" ], "model_class": "video generation", "modalities_in": [ "text", "image", "video" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Alibaba Cloud PAI releases InP (start and end frame) and Control variants; Wan2.2-Fun-A14B-Control is fine-tuned from Wan2.2-I2V-A14B.", "context_length": null, "release_date": "2025-03-17", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Alibaba Cloud PAI variants of Wan for frame inpainting and control-guided video generation.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/alibaba-pai/Wan2.1-Fun-1.3B-InP", "https://www.modelscope.cn/models/PAI/Wan2.1-Fun-1.3B-InP" ], "unverified": [] }, { "id": "alibaba-modelscope-t2v", "name": "ModelScope Text-to-Video (ModelScopeT2V)", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": null, "params_total": 1411233860, "params_active": null, "params_checkpoint": null, "params_note": "1.7B in the repository name. Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2023-03-22", "release_date_basis": "registry date", "weights_date": "2023-03-22", "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Early open text-to-video diffusion model from Alibaba DAMO Academy.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2308.06571", "https://huggingface.co/ali-vilab/text-to-video-ms-1.7b", "https://huggingface.co/api/models?pipeline_tag=text-to-video&sort=likes&limit=50", "https://www.modelscope.cn/models/iic/text-to-video-synthesis" ], "unverified": [ "release_date" ] }, { "id": "alibaba-i2vgen-xl", "name": "I2VGen-XL", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "video generation", "modalities_in": [ "image", "text" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": null, "params_total": 1420469224, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2023-11-07", "release_date_basis": "announcement page", "weights_date": "2023-12-15", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Cascaded diffusion model from Alibaba that turns an image and text into high-quality video.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2311.04145", "https://huggingface.co/ali-vilab/i2vgen-xl", "https://www.modelscope.cn/models/iic/Image-to-Video" ], "unverified": [] }, { "id": "alibaba-unianimate", "name": "UniAnimate", "level": "model", "family_id": null, "developer": [ "alibaba" ], "model_class": "avatar", "modalities_in": [ "image", "video" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-06-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Unified video diffusion model that animates a human image to follow a pose sequence consistently.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2406.01188", "https://www.modelscope.cn/models/iic/UniAnimate" ], "unverified": [] }, { "id": "openmoss-mova", "name": "MOVA", "level": "model", "family_id": null, "developer": [ "openmoss" ], "model_class": "video generation", "modalities_in": [ "image", "text" ], "modalities_out": [ "video", "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "360p and 720p checkpoints.", "context_length": null, "release_date": "2026-01-28", "release_date_basis": "registry date", "weights_date": "2026-01-28", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open model that generates video with synchronised audio from an image and text.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2602.08794", "https://huggingface.co/OpenMOSS-Team/MOVA-360p", "https://huggingface.co/api/models?pipeline_tag=image-to-video&sort=downloads&limit=50" ], "unverified": [ "release_date" ] }, { "id": "createai-ruyi-mini-7b", "name": "Ruyi-Mini-7B", "level": "model", "family_id": null, "developer": [], "model_class": "video generation", "modalities_in": [ "image" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": 7244452992, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2024-12-16", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Open 7B image-to-video model released by CreateAI.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/IamCreateAI/Ruyi-Mini-7B", "https://huggingface.co/api/models?pipeline_tag=image-to-video&sort=likes&limit=50" ], "unverified": [] }, { "id": "bytedance-bernini-r", "name": "Bernini-R", "level": "model", "family_id": null, "developer": [ "bytedance" ], "model_class": "video generation", "modalities_in": [ "image", "text" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-05-21", "release_date_basis": "announcement page", "weights_date": "2026-06-01", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ByteDance image-and-text to video diffusion model that plans in a latent semantic space before generating.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2605.22344", "https://huggingface.co/ByteDance/Bernini-R", "https://huggingface.co/api/models?pipeline_tag=image-text-to-video&sort=likes&limit=50" ], "unverified": [] }, { "id": "animatediff-family", "name": "AnimateDiff", "level": "family", "family_id": null, "developer": [ "shanghai-ai-lab", "cuhk", "stanford-university" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-07-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Motion modules that animate personalised text-to-image diffusion models into short videos without model-specific tuning.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2307.04725", "https://github.com/open-mmlab/mmagic/tree/main/configs/animatediff", "https://arxiv.org/html/2307.04725" ], "unverified": [] }, { "id": "animatediff", "name": "AnimateDiff", "level": "variant", "family_id": "animatediff-family", "developer": [ "shanghai-ai-lab", "cuhk", "stanford-university" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-07-10", "release_date_basis": "announcement page", "weights_date": "2023-07-18", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Original AnimateDiff motion module for Stable Diffusion models.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2307.04725", "https://huggingface.co/guoyww/animatediff", "https://github.com/open-mmlab/mmagic/tree/main/configs/animatediff", "https://replicate.com/lucataco/animate-diff", "https://arxiv.org/html/2307.04725" ], "unverified": [] }, { "id": "bytedance-animatediff-lightning", "name": "AnimateDiff-Lightning", "level": "variant", "family_id": "animatediff-family", "developer": [ "bytedance" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "CreativeML OpenRAIL-M licence tag.", "context_length": null, "release_date": "2024-03-19", "release_date_basis": "announcement page", "weights_date": "2024-03-19", "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ByteDance few-step distilled AnimateDiff for fast text-to-video.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2403.12706", "https://huggingface.co/ByteDance/AnimateDiff-Lightning", "https://huggingface.co/api/models?pipeline_tag=text-to-video&sort=likes&limit=50" ], "unverified": [] }, { "id": "animatelcm", "name": "AnimateLCM", "level": "variant", "family_id": "animatediff-family", "developer": [ "cuhk", "sensetime" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": null, "params_total": 453813440, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2024-02-01", "release_date_basis": "announcement page", "weights_date": "2024-02-03", "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Consistency-distilled animation model for fast personalised video generation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2402.00769", "https://huggingface.co/wangfuyun/AnimateLCM", "https://huggingface.co/api/models?pipeline_tag=text-to-video&sort=likes&limit=50", "https://arxiv.org/html/2402.00769" ], "unverified": [] }, { "id": "pia-image-animator", "name": "PIA (Personalized Image Animator)", "level": "model", "family_id": null, "developer": [ "shanghai-ai-lab" ], "model_class": "video generation", "modalities_in": [ "image", "text" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-12-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Plug-and-play module that animates a still image under text control in personalised text-to-image models.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2312.13964", "https://github.com/open-mmlab/PIA", "https://arxiv.org/html/2312.13964" ], "unverified": [] }, { "id": "live2diff", "name": "Live2Diff", "level": "model", "family_id": null, "developer": [ "shanghai-ai-lab" ], "model_class": "video generation", "modalities_in": [ "video" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-07-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Video diffusion model with uni-directional temporal attention for live-stream video translation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2407.08701", "https://github.com/open-mmlab/Live2Diff", "https://arxiv.org/html/2407.08701" ], "unverified": [] }, { "id": "genmo-mochi-1", "name": "Mochi 1", "level": "model", "family_id": null, "developer": [ "genmo" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": null, "params_total": 10027677744, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2024-10-22", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Genmo open text-to-video diffusion model released as a preview.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/genmo/mochi-1-preview", "https://huggingface.co/api/models?pipeline_tag=text-to-video&sort=likes&limit=50", "https://replicate.com/genmoai/mochi-1" ], "unverified": [] }, { "id": "zeroscope-v2", "name": "Zeroscope v2", "level": "model", "family_id": null, "developer": [], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-06-21", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Community open text-to-video model producing 576px clips.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/cerspense/zeroscope_v2_576w", "https://huggingface.co/api/models?pipeline_tag=text-to-video&sort=likes&limit=50" ], "unverified": [] }, { "id": "stepfun-step-video-t2v", "name": "Step-Video-T2V", "level": "model", "family_id": null, "developer": [ "stepfun" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": null, "params_total": 29342478400, "params_active": null, "params_checkpoint": null, "params_note": "A turbo variant (stepvideo-t2v-turbo) is also published. Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2025-02-14", "release_date_basis": "announcement page", "weights_date": "2025-02-14", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [ "multi-GPU node" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "StepFun open text-to-video foundation model of about 30B parameters.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2502.10248", "https://huggingface.co/stepfun-ai/stepvideo-t2v", "https://huggingface.co/api/models?pipeline_tag=text-to-video&sort=likes&limit=50", "https://www.modelscope.cn/models/stepfun-ai/stepvideo-t2v" ], "unverified": [] }, { "id": "bytedance-seedvr-family", "name": "SeedVR", "level": "family", "family_id": null, "developer": [ "bytedance" ], "model_class": "video generation", "modalities_in": [ "video" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ByteDance Seed diffusion transformers for generic video restoration and upscaling.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/api/models?pipeline_tag=video-to-video&sort=downloads&limit=50" ], "unverified": [] }, { "id": "bytedance-seedvr", "name": "SeedVR", "level": "model", "family_id": "bytedance-seedvr-family", "developer": [ "bytedance" ], "model_class": "video generation", "modalities_in": [ "video" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-01-02", "release_date_basis": "announcement page", "weights_date": "2025-06-11", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Diffusion transformer for generic video restoration at any resolution.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2501.01320", "https://huggingface.co/ByteDance-Seed/SeedVR-7B" ], "unverified": [] }, { "id": "bytedance-seedvr2", "name": "SeedVR2", "level": "model", "family_id": "bytedance-seedvr-family", "developer": [ "bytedance" ], "model_class": "video generation", "modalities_in": [ "video" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "3B and 7B checkpoints.", "context_length": null, "release_date": "2025-06-05", "release_date_basis": "announcement page", "weights_date": "2025-06-10", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "One-step video restoration model trained with diffusion adversarial post-training.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2506.05301", "https://huggingface.co/ByteDance-Seed/SeedVR2-3B", "https://huggingface.co/api/models?pipeline_tag=video-to-video&sort=downloads&limit=50" ], "unverified": [] }, { "id": "flashvsr", "name": "FlashVSR", "level": "model", "family_id": null, "developer": [ "tsinghua-university" ], "model_class": "video generation", "modalities_in": [ "video" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-10-14", "release_date_basis": "announcement page", "weights_date": "2025-10-31", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Diffusion-based streaming video super-resolution aimed at real-time use.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2510.12747", "https://huggingface.co/JunhaoZhuang/FlashVSR-v1.1", "https://huggingface.co/api/models?pipeline_tag=video-to-video&sort=downloads&limit=50", "https://arxiv.org/html/2510.12747" ], "unverified": [] }, { "id": "decart-lucy-edit-dev", "name": "Lucy Edit Dev", "level": "model", "family_id": null, "developer": [ "decart" ], "model_class": "video generation", "modalities_in": [ "video", "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": 5000377536, "params_active": null, "params_checkpoint": null, "params_note": "Lucy Edit Dev non-commercial licence. Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2025-09-16", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Decart open video editing model that edits videos from text instructions.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/decart-ai/Lucy-Edit-Dev", "https://huggingface.co/api/models?pipeline_tag=video-to-video&sort=likes&limit=50" ], "unverified": [] }, { "id": "netflix-void", "name": "VOID", "level": "model", "family_id": null, "developer": [ "netflix" ], "model_class": "video generation", "modalities_in": [ "video" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-03-30", "release_date_basis": "registry date", "weights_date": "2026-03-30", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Netflix model that deletes objects from video together with their interactions, such as shadows and reflections.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2604.02296", "https://huggingface.co/netflix/void-model", "https://huggingface.co/api/models?pipeline_tag=video-to-video&sort=likes&limit=50" ], "unverified": [] }, { "id": "skywork-ai-skyreels-v2", "name": "SkyReels-V2", "level": "model", "family_id": null, "developer": [ "skywork-ai" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": null, "params_total": 14288491584, "params_active": null, "params_checkpoint": null, "params_note": "Skywork licence. Earlier than SkyReels V4 held in AURION. Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2025-04-17", "release_date_basis": "announcement page", "weights_date": "2025-04-18", "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Skywork open film generative model for infinite-length video.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2504.13074", "https://huggingface.co/Skywork/SkyReels-V2-T2V-14B-720P", "https://www.modelscope.cn/models/Skywork/SkyReels-V2-T2V-14B-720P" ], "unverified": [] }, { "id": "character-ai-ovi", "name": "Ovi", "level": "model", "family_id": null, "developer": [ "character-ai" ], "model_class": "video generation", "modalities_in": [ "image", "text" ], "modalities_out": [ "video", "audio" ], "architecture": "diffusion", "architecture_detail": "Twin backbone cross-modal fusion", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-09-30", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Model that generates video with synchronised audio from text or an image and text, using twin audio and video backbones.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2510.01284", "https://replicate.com/character-ai/ovi-i2v" ], "unverified": [] }, { "id": "lightricks-ltx-2-3", "name": "LTX-2.3", "level": "model", "family_id": "lightricks-ltx", "developer": [ "lightricks" ], "model_class": "video generation", "modalities_in": [ "text", "image", "audio" ], "modalities_out": [ "video", "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-03-04", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "ltx-2-community", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": "Up to 4K at 50 fps (Replicate page).", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Lightricks joint audio-video model with portrait support, audio-to-video, retake and extend.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/Lightricks/LTX-2.3", "https://arxiv.org/abs/2601.03233", "https://replicate.com/lightricks/ltx-2.3-pro" ], "unverified": [] }, { "id": "luma-ai-ray-2", "name": "Ray2", "level": "model", "family_id": null, "developer": [ "luma-ai" ], "model_class": "video generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-01-23", "release_date_basis": "registry date", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": "Bedrock model ID luma.ray-v2:0 in us-west-2; 5 s and 9 s clips at 720p.", "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "cross-region only", "status": "active", "summary": "Luma AI large-scale video generation model for realistic clips with coherent motion from text and images.", "verified_on": "2026-10-12", "sources": [ "https://docs.aws.amazon.com/bedrock/latest/userguide/model-card-luma-ai-ray2.html", "https://replicate.com/luma/ray-2-720p" ], "unverified": [] }, { "id": "pyramid-flow", "name": "Pyramid Flow", "level": "model", "family_id": null, "developer": [ "peking-university", "kuaishou" ], "model_class": "video generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "video" ], "architecture": "flow matching", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "An SD3-based checkpoint (pyramid-flow-sd3) carries the Stability AI community licence.", "context_length": null, "release_date": "2024-10-08", "release_date_basis": "announcement page", "weights_date": "2024-10-29", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Autoregressive video generation model that uses pyramidal flow matching for efficiency.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2410.05954", "https://huggingface.co/rain1011/pyramid-flow-miniflux", "https://replicate.com/zsxkib/pyramid-flow", "https://arxiv.org/html/2410.05954" ], "unverified": [] }, { "id": "google-film", "name": "FILM (Frame Interpolation for Large Motion)", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "video generation", "modalities_in": [ "image" ], "modalities_out": [ "video" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-02-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Frame interpolation network that synthesises in-between frames for large scene motion.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2202.04901", "https://www.kaggle.com/models/google/film" ], "unverified": [] }, { "id": "toflow", "name": "TOFlow", "level": "model", "family_id": null, "developer": [ "massachusetts-institute-of-technology", "google" ], "model_class": "video generation", "modalities_in": [ "video" ], "modalities_out": [ "video" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2017-11-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Video enhancement with task-oriented optical flow for interpolation, denoising and super-resolution.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1711.09078", "https://github.com/open-mmlab/mmagic/tree/main/configs/tof", "https://arxiv.org/html/1711.09078" ], "unverified": [] }, { "id": "flavr", "name": "FLAVR", "level": "model", "family_id": null, "developer": [ "uc-san-diego", "carnegie-mellon-university" ], "model_class": "video generation", "modalities_in": [ "video" ], "modalities_out": [ "video" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2020-12-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Flow-agnostic 3D convolution model for fast video frame interpolation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2012.08512", "https://github.com/open-mmlab/mmagic/tree/main/configs/flavr", "https://arxiv.org/html/2012.08512" ], "unverified": [] }, { "id": "edvr", "name": "EDVR", "level": "model", "family_id": null, "developer": [ "cuhk", "ntu-singapore" ], "model_class": "video generation", "modalities_in": [ "video" ], "modalities_out": [ "video" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2019-05-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Video restoration with enhanced deformable convolutions for super-resolution and deblurring.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1905.02716", "https://github.com/open-mmlab/mmagic/tree/main/configs/edvr", "https://arxiv.org/html/1905.02716" ], "unverified": [] }, { "id": "tdan", "name": "TDAN", "level": "model", "family_id": null, "developer": [ "university-of-rochester" ], "model_class": "video generation", "modalities_in": [ "video" ], "modalities_out": [ "video" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2018-12-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Temporally deformable alignment network for video super-resolution.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1812.02898", "https://github.com/open-mmlab/mmagic/tree/main/configs/tdan", "https://arxiv.org/html/1812.02898" ], "unverified": [] }, { "id": "basicvsr-family", "name": "BasicVSR", "level": "family", "family_id": null, "developer": [ "ntu-singapore" ], "model_class": "video generation", "modalities_in": [ "video" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Video super-resolution line: BasicVSR and IconVSR, BasicVSR++ and RealBasicVSR.", "verified_on": "2026-10-12", "sources": [ "https://github.com/open-mmlab/mmagic/tree/main/configs/basicvsr", "https://arxiv.org/html/2012.02181" ], "unverified": [] }, { "id": "basicvsr", "name": "BasicVSR and IconVSR", "level": "variant", "family_id": "basicvsr-family", "developer": [ "ntu-singapore" ], "model_class": "video generation", "modalities_in": [ "video" ], "modalities_out": [ "video" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2020-12-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Bidirectional propagation baseline for video super-resolution.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2012.02181", "https://github.com/open-mmlab/mmagic/tree/main/configs/basicvsr", "https://arxiv.org/html/2012.02181" ], "unverified": [] }, { "id": "basicvsr-plus-plus", "name": "BasicVSR++", "level": "variant", "family_id": "basicvsr-family", "developer": [ "ntu-singapore" ], "model_class": "video generation", "modalities_in": [ "video" ], "modalities_out": [ "video" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2021-04-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Video super-resolution with second-order grid propagation and flow-guided deformable alignment.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2104.13371", "https://github.com/open-mmlab/mmagic/tree/main/configs/basicvsr_pp", "https://arxiv.org/html/2104.13371" ], "unverified": [] }, { "id": "realbasicvsr", "name": "RealBasicVSR", "level": "variant", "family_id": "basicvsr-family", "developer": [ "ntu-singapore" ], "model_class": "video generation", "modalities_in": [ "video" ], "modalities_out": [ "video" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMagic; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2021-11-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Video super-resolution for real-world degradations.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2111.12704", "https://github.com/open-mmlab/mmagic/tree/main/configs/real_basicvsr", "https://arxiv.org/html/2111.12704" ], "unverified": [] }, { "id": "nvidia-audio2face-3d", "name": "Audio2Face-3D", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "avatar", "modalities_in": [ "audio" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-06-27", "release_date_basis": "registry date", "weights_date": "2025-06-27", "openness": "open weights restricted", "licence": "nvidia-open-model-license", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA model that drives realistic 3D facial animation of digital avatars from speech audio.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2508.16401", "https://huggingface.co/nvidia/Audio2Face-3D-v3.0", "https://catalog.ngc.nvidia.com/orgs/nim/teams/nvidia/models/audio2emotion_model" ], "unverified": [ "release_date" ] }, { "id": "nvidia-audio2emotion", "name": "Audio2Emotion", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "avatar", "modalities_in": [ "audio" ], "modalities_out": [ "text" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "NVIDIA Audio2Emotion licence; gated.", "context_length": null, "release_date": "2025-08-01", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NVIDIA model that infers emotion from speech audio to steer Audio2Face-3D facial animation.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/nvidia/Audio2Emotion-v2.2", "https://arxiv.org/abs/2508.16401", "https://catalog.ngc.nvidia.com/orgs/nim/teams/nvidia/models/audio2emotion_model" ], "unverified": [] }, { "id": "faceshot", "name": "FaceShot", "level": "model", "family_id": null, "developer": [ "tongji-university", "shanghai-ai-lab" ], "model_class": "avatar", "modalities_in": [ "image", "video" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Training-free method that animates the face of any character, including non-human ones, from a driving video.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2503.00740", "https://github.com/open-mmlab/FaceShot", "https://arxiv.org/html/2503.00740" ], "unverified": [] }, { "id": "bytedance-omnihuman", "name": "OmniHuman", "level": "family", "family_id": null, "developer": [ "bytedance" ], "model_class": "avatar", "modalities_in": [ "image", "audio", "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ByteDance human animation models that generate realistic video from a single image and audio.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2502.01061", "https://replicate.com/bytedance/omni-human-1.5" ], "unverified": [] }, { "id": "bytedance-omnihuman-1", "name": "OmniHuman-1", "level": "model", "family_id": "bytedance-omnihuman", "developer": [ "bytedance" ], "model_class": "avatar", "modalities_in": [ "image", "audio" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-02-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "One-stage conditioned human animation model that turns an image plus audio or video signals into human video.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2502.01061", "https://replicate.com/bytedance/omni-human" ], "unverified": [] }, { "id": "bytedance-omnihuman-1-5", "name": "OmniHuman-1.5", "level": "model", "family_id": "bytedance-omnihuman", "developer": [ "bytedance" ], "model_class": "avatar", "modalities_in": [ "image", "audio", "text" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-08-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Digital human model that generates film-grade video from one image, an audio clip and an optional text prompt.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2508.19209", "https://replicate.com/bytedance/omni-human-1.5" ], "unverified": [] }, { "id": "bytedance-dreamactor", "name": "DreamActor", "level": "family", "family_id": null, "developer": [ "bytedance" ], "model_class": "avatar", "modalities_in": [ "image", "video" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "ByteDance human image animation models driven by a reference video.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2504.01724", "https://replicate.com/bytedance/dreamactor-m2.0" ], "unverified": [] }, { "id": "bytedance-dreamactor-m1", "name": "DreamActor-M1", "level": "model", "family_id": "bytedance-dreamactor", "developer": [ "bytedance" ], "model_class": "avatar", "modalities_in": [ "image", "video" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-04-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [ "research only" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Holistic human image animation with hybrid guidance for face, body and appearance.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2504.01724" ], "unverified": [] }, { "id": "sadtalker", "name": "SadTalker", "level": "model", "family_id": null, "developer": [ "xi-an-jiaotong-university", "tencent" ], "model_class": "avatar", "modalities_in": [ "image", "audio" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2022-11-22", "release_date_basis": "announcement page", "weights_date": "2023-04-08", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Talking-head generator that predicts 3D motion coefficients from audio to animate a single face image.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2211.12194", "https://huggingface.co/vinthony/SadTalker", "https://replicate.com/cjwbw/sadtalker", "https://arxiv.org/html/2211.12194" ], "unverified": [] }, { "id": "meituan-multitalk", "name": "MultiTalk", "level": "model", "family_id": null, "developer": [ "meituan" ], "model_class": "avatar", "modalities_in": [ "image", "audio", "text" ], "modalities_out": [ "video" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-05-28", "release_date_basis": "announcement page", "weights_date": "2025-06-09", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Audio-driven model that generates conversational video with multiple speaking people.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2505.22647", "https://huggingface.co/MeiGen-AI/MeiGen-MultiTalk", "https://replicate.com/zsxkib/multitalk" ], "unverified": [] }, { "id": "first-order-motion-model", "name": "First Order Motion Model", "level": "model", "family_id": null, "developer": [ "university-of-trento", "snap" ], "model_class": "avatar", "modalities_in": [ "image", "video" ], "modalities_out": [ "video" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Qualcomm AI Hub ships a build (fomm); repository licence MIT per the coverage entry.", "context_length": null, "release_date": "2020-02-29", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Image animation model that transfers motion from a driving video to a still image of an object or face.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2003.00196", "https://aihub.qualcomm.com/models/fomm", "https://github.com/AliaksandrSiarohin/first-order-model/tree/master", "https://arxiv.org/html/2003.00196" ], "unverified": [] }, { "id": "microsoft-trellis", "name": "TRELLIS (image-large)", "level": "model", "family_id": null, "developer": [ "microsoft" ], "model_class": "3D", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Text-conditioned TRELLIS checkpoints also exist. AURION holds the later TRELLIS.2 4B.", "context_length": null, "release_date": "2024-12-02", "release_date_basis": "announcement page", "weights_date": "2024-12-02", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Microsoft 3D generator that uses structured 3D latents to create 3D assets from images or text.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2412.01506", "https://huggingface.co/microsoft/TRELLIS-image-large", "https://huggingface.co/api/models?pipeline_tag=image-to-3d&sort=downloads&limit=50", "https://www.modelscope.cn/models/AI-ModelScope/TRELLIS-image-large" ], "unverified": [] }, { "id": "meta-vggt-1b", "name": "VGGT", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "3D", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 1256537516, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2025-03-11", "release_date_basis": "registry date", "weights_date": "2025-03-11", "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Feed-forward transformer that infers camera parameters, depth, point maps and tracks from one or many views of a scene.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2503.11651", "https://huggingface.co/facebook/VGGT-1B", "https://huggingface.co/api/models?pipeline_tag=image-to-3d&sort=downloads&limit=50", "https://www.modelscope.cn/models/facebook/VGGT-1B" ], "unverified": [] }, { "id": "meta-mapanything", "name": "MapAnything", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "3D", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": 1228492758, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2025-09-08", "release_date_basis": "registry date", "weights_date": "2025-09-08", "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Universal feed-forward model for metric 3D reconstruction from images and optional geometric inputs.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2509.13414", "https://huggingface.co/facebook/map-anything", "https://huggingface.co/api/models?pipeline_tag=image-to-3d&sort=downloads&limit=50" ], "unverified": [ "release_date" ] }, { "id": "pi3", "name": "Pi3 and Pi3X", "level": "model", "family_id": null, "developer": [ "shanghai-ai-lab" ], "model_class": "3D", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": 1360032439, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2025-07-17", "release_date_basis": "announcement page", "weights_date": "2025-12-27", "openness": "open weights restricted", "licence": "cc-by-nc-4-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Permutation-equivariant feed-forward model for visual geometry and 3D reconstruction without a fixed reference view.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2507.13347", "https://huggingface.co/yyfz233/Pi3X", "https://huggingface.co/api/models?pipeline_tag=image-to-3d&sort=downloads&limit=50", "https://arxiv.org/html/2507.13347" ], "unverified": [] }, { "id": "naver-dust3r", "name": "DUSt3R", "level": "model", "family_id": null, "developer": [ "naver" ], "model_class": "3D", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": 571171208, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2023-12-21", "release_date_basis": "announcement page", "weights_date": "2024-06-24", "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "NAVER LABS model that regresses dense 3D point maps from image pairs without known camera poses.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2312.14132", "https://huggingface.co/naver/DUSt3R_ViTLarge_BaseDecoder_512_dpt", "https://huggingface.co/api/models?pipeline_tag=image-to-3d&sort=downloads&limit=50" ], "unverified": [] }, { "id": "naver-mast3r", "name": "MASt3R", "level": "model", "family_id": null, "developer": [ "naver" ], "model_class": "3D", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": 688638856, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2024-06-14", "release_date_basis": "announcement page", "weights_date": "2024-07-09", "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Extension of DUSt3R that grounds image matching in 3D with metric point maps and dense local features.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2406.09756", "https://huggingface.co/naver/MASt3R_ViTLarge_BaseDecoder_512_catmlpdpt_metric", "https://huggingface.co/api/models?pipeline_tag=image-to-3d&sort=downloads&limit=50" ], "unverified": [] }, { "id": "tencent-hunyuan3d", "name": "Hunyuan3D", "level": "family", "family_id": null, "developer": [ "tencent" ], "model_class": "3D", "modalities_in": [ "image", "text" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Hunyuan3D 2.1 is held as its own record (tencent-hunyuan3d-2-1).", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tencent model line for generating textured 3D assets from images and text.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/api/models?pipeline_tag=image-to-3d&sort=downloads&limit=50", "https://replicate.com/tencent/hunyuan-3d-3.1" ], "unverified": [] }, { "id": "tencent-hunyuan3d-1-0", "name": "Hunyuan3D 1.0", "level": "variant", "family_id": "tencent-hunyuan3d", "developer": [ "tencent" ], "model_class": "3D", "modalities_in": [ "image", "text" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-11-01", "release_date_basis": "registry date", "weights_date": "2024-11-01", "openness": "open weights restricted", "licence": "tencent-hunyuan-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Unified framework for text-to-3D and image-to-3D generation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2411.02293", "https://huggingface.co/tencent/Hunyuan3D-1", "https://huggingface.co/api/models?pipeline_tag=image-to-3d&sort=downloads&limit=50" ], "unverified": [] }, { "id": "tencent-hunyuan3d-2-0", "name": "Hunyuan3D 2.0", "level": "variant", "family_id": "tencent-hunyuan3d", "developer": [ "tencent" ], "model_class": "3D", "modalities_in": [ "image", "text" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-01-20", "release_date_basis": "registry date", "weights_date": "2025-01-20", "openness": "open weights restricted", "licence": "tencent-hunyuan-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Scaled diffusion models for high-resolution textured 3D asset generation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2501.12202", "https://huggingface.co/tencent/Hunyuan3D-2", "https://huggingface.co/api/models?pipeline_tag=image-to-3d&sort=downloads&limit=50" ], "unverified": [] }, { "id": "tencent-hunyuan3d-2-mini", "name": "Hunyuan3D 2mini", "level": "variant", "family_id": "tencent-hunyuan3d", "developer": [ "tencent" ], "model_class": "3D", "modalities_in": [ "image", "text" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "tencent-hunyuan-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Smaller Hunyuan3D 2 shape model.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/tencent/Hunyuan3D-2mini", "https://huggingface.co/api/models?pipeline_tag=image-to-3d&sort=downloads&limit=50" ], "unverified": [] }, { "id": "tencent-hunyuan3d-2-mv", "name": "Hunyuan3D 2mv", "level": "variant", "family_id": "tencent-hunyuan3d", "developer": [ "tencent" ], "model_class": "3D", "modalities_in": [ "image", "text" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-03-12", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": "tencent-hunyuan-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multi-view conditioned Hunyuan3D 2 shape model.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/tencent/Hunyuan3D-2mv", "https://huggingface.co/api/models?pipeline_tag=image-to-3d&sort=downloads&limit=50" ], "unverified": [] }, { "id": "tencent-hunyuan3d-omni", "name": "Hunyuan3D-Omni", "level": "variant", "family_id": "tencent-hunyuan3d", "developer": [ "tencent" ], "model_class": "3D", "modalities_in": [ "image", "text" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-09-23", "release_date_basis": "registry date", "weights_date": "2025-09-23", "openness": "open weights restricted", "licence": "tencent-hunyuan-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Hunyuan3D model for controllable 3D asset generation from images plus extra controls.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2509.21245", "https://huggingface.co/tencent/Hunyuan3D-Omni", "https://huggingface.co/api/models?pipeline_tag=image-to-3d&sort=downloads&limit=50" ], "unverified": [] }, { "id": "tencent-hunyuanworld-mirror", "name": "HunyuanWorld-Mirror", "level": "model", "family_id": null, "developer": [ "tencent" ], "model_class": "3D", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": 1263337002, "params_active": null, "params_checkpoint": null, "params_note": "HunyuanWorld-Mirror community licence. Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2025-10-12", "release_date_basis": "announcement page", "weights_date": "2025-10-16", "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Feed-forward model for universal 3D world reconstruction that accepts any available priors such as poses or depth.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2510.10726", "https://huggingface.co/tencent/HunyuanWorld-Mirror", "https://huggingface.co/api/models?pipeline_tag=image-to-3d&sort=likes&limit=50", "https://www.modelscope.cn/models/Tencent-Hunyuan/HunyuanWorld-Mirror" ], "unverified": [] }, { "id": "tencent-instantmesh", "name": "InstantMesh", "level": "model", "family_id": null, "developer": [ "tencent" ], "model_class": "3D", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-04-10", "release_date_basis": "announcement page", "weights_date": "2024-04-10", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tencent ARC model that generates a 3D mesh from a single image via sparse-view reconstruction.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2404.07191", "https://huggingface.co/TencentARC/InstantMesh", "https://huggingface.co/api/models?pipeline_tag=image-to-3d&sort=downloads&limit=50" ], "unverified": [] }, { "id": "tencent-pixal3d", "name": "Pixal3D", "level": "model", "family_id": null, "developer": [ "tencent" ], "model_class": "3D", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-04-30", "release_date_basis": "registry date", "weights_date": "2026-04-30", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tencent ARC pixel-aligned 3D generation from images.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2605.10922", "https://huggingface.co/TencentARC/Pixal3D", "https://huggingface.co/api/models?pipeline_tag=image-to-3d&sort=likes&limit=50" ], "unverified": [ "release_date" ] }, { "id": "stability-ai-stable-zero123", "name": "Stable Zero123", "level": "model", "family_id": null, "developer": [ "stability-ai" ], "model_class": "3D", "modalities_in": [ "image" ], "modalities_out": [ "image" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Stability AI non-commercial community licence.", "context_length": null, "release_date": "2023-12-08", "release_date_basis": "registry date", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Stability AI model that generates novel views of an object from one image, used for image-to-3D.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/stabilityai/stable-zero123", "https://huggingface.co/api/models?pipeline_tag=image-to-3d&sort=downloads&limit=50" ], "unverified": [] }, { "id": "stability-ai-sv3d", "name": "SV3D", "level": "model", "family_id": "stability-ai-stable-video-diffusion", "developer": [ "stability-ai" ], "model_class": "3D", "modalities_in": [ "image" ], "modalities_out": [ "video", "3D" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "SV3D non-commercial community licence; gated.", "context_length": null, "release_date": "2024-03-18", "release_date_basis": "announcement page", "weights_date": "2024-03-18", "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Latent video diffusion model that generates orbital multi-view videos of an object from one image for 3D generation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2403.12008", "https://huggingface.co/stabilityai/sv3d", "https://huggingface.co/api/models?pipeline_tag=image-to-video&sort=downloads&limit=50" ], "unverified": [] }, { "id": "apple-sharp", "name": "SHARP", "level": "model", "family_id": null, "developer": [ "apple" ], "model_class": "3D", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Apple AMLR licence tag.", "context_length": null, "release_date": "2025-12-11", "release_date_basis": "announcement page", "weights_date": "2025-12-12", "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Apple model for monocular view synthesis that builds a 3D Gaussian representation from one image in under a second.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2512.10685", "https://huggingface.co/api/models?pipeline_tag=image-to-3d&sort=likes&limit=50", "https://huggingface.co/apple/Sharp" ], "unverified": [] }, { "id": "openai-shap-e", "name": "Shap-E", "level": "model", "family_id": null, "developer": [ "openai" ], "model_class": "3D", "modalities_in": [ "text", "image" ], "modalities_out": [ "3D" ], "architecture": "diffusion", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-05-03", "release_date_basis": "announcement page", "weights_date": "2023-07-04", "openness": "open weights permissive", "licence": "mit", "access_routes": [ "open weights download", "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "OpenAI conditional generator of 3D implicit functions that can be rendered as meshes or radiance fields from text or images.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2305.02463", "https://huggingface.co/openai/shap-e", "https://huggingface.co/api/models?pipeline_tag=text-to-3d&sort=downloads&limit=50", "https://replicate.com/cjwbw/shap-e" ], "unverified": [] }, { "id": "llama-mesh", "name": "LLaMA-Mesh", "level": "model", "family_id": null, "developer": [ "tsinghua-university", "nvidia" ], "model_class": "3D", "modalities_in": [ "text" ], "modalities_out": [ "text", "3D" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": 8030261248, "params_active": null, "params_checkpoint": null, "params_note": "Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2024-11-14", "release_date_basis": "announcement page", "weights_date": "2024-11-17", "openness": "open weights restricted", "licence": "llama-3-1-community", "access_routes": [ "open weights download", "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Language model fine-tuned to generate 3D meshes as plain text, unifying mesh generation with chat.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2411.09595", "https://huggingface.co/Zhengyi/LLaMA-Mesh", "https://huggingface.co/api/models?pipeline_tag=text-to-3d&sort=downloads&limit=50", "https://ai.azure.com/catalog/models?task=text-to-3d", "https://arxiv.org/html/2411.09595" ], "unverified": [] }, { "id": "tencent-hy-motion-1-0", "name": "HY-Motion 1.0", "level": "model", "family_id": null, "developer": [ "tencent" ], "model_class": "3D", "modalities_in": [ "text" ], "modalities_out": [ "3D" ], "architecture": "flow matching", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-12-29", "release_date_basis": "announcement page", "weights_date": "2025-12-29", "openness": "open weights restricted", "licence": "tencent-hunyuan-community", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tencent flow matching model that generates 3D human motion from text.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2512.23464", "https://huggingface.co/tencent/HY-Motion-1.0", "https://huggingface.co/api/models?pipeline_tag=text-to-3d&sort=downloads&limit=50" ], "unverified": [] }, { "id": "intel-ldm3d", "name": "LDM3D", "level": "model", "family_id": null, "developer": [ "intel" ], "model_class": "3D", "modalities_in": [ "text" ], "modalities_out": [ "image", "3D" ], "architecture": "diffusion", "architecture_detail": null, "params_total": 859520964, "params_active": null, "params_checkpoint": null, "params_note": "CreativeML OpenRAIL-M licence tag; ldm3d and ldm3d-4c checkpoints. Parameter count from the Hugging Face safetensors total.", "context_length": null, "release_date": "2023-05-18", "release_date_basis": "announcement page", "weights_date": "2023-06-22", "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Intel latent diffusion model that generates an RGB image and its depth map from text.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2305.10853", "https://huggingface.co/Intel/ldm3d-4c", "https://huggingface.co/api/models?pipeline_tag=text-to-3d&sort=downloads&limit=50" ], "unverified": [] }, { "id": "nvidia-eg3d", "name": "EG3D", "level": "model", "family_id": null, "developer": [ "nvidia" ], "model_class": "3D", "modalities_in": [], "modalities_out": [ "image", "3D" ], "architecture": "CNN", "architecture_detail": "Tri-plane 3D-aware GAN", "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2021-12-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Efficient geometry-aware 3D GAN that generates multi-view consistent images and shapes using a tri-plane representation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2112.07945", "https://catalog.ngc.nvidia.com/orgs/nvidia/teams/research/models/eg3d", "https://github.com/open-mmlab/mmagic/tree/main/configs/eg3d" ], "unverified": [] }, { "id": "google-zip-nerf", "name": "Zip-NeRF", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "3D", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-04-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Anti-aliased grid-based neural radiance field for reconstructing 3D scenes from photos; offered as a Vertex AI Model Garden notebook.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2304.06706", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden" ], "unverified": [] }, { "id": "google-camp", "name": "CamP", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "3D", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-08-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "cloud marketplace" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Camera preconditioning method that improves joint camera and NeRF optimisation; offered as a Vertex AI Model Garden notebook.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2308.10902", "https://github.com/GoogleCloudPlatform/vertex-ai-samples/tree/main/notebooks/community/model_garden" ], "unverified": [] }, { "id": "meta-mesh-rcnn", "name": "Mesh R-CNN", "level": "model", "family_id": null, "developer": [ "meta" ], "model_class": "3D", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Model zoo weights CC BY-SA 3.0, code Apache-2.0 (coverage entry).", "context_length": null, "release_date": "2019-06-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Detects objects in an image and predicts a 3D triangle mesh for each.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1906.02739", "https://github.com/facebookresearch/meshrcnn" ], "unverified": [] }, { "id": "toyota-research-institute-dd3d", "name": "DD3D", "level": "model", "family_id": null, "developer": [ "toyota-research-institute" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Model zoo weights CC BY-SA 3.0, code Apache-2.0 (coverage entry).", "context_length": null, "release_date": "2021-08-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Monocular 3D object detector with dense depth pre-training, without a pseudo-LiDAR step.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2108.06417", "https://github.com/TRI-ML/dd3d" ], "unverified": [] }, { "id": "google-objectron", "name": "Objectron 3D object detection", "level": "model", "family_id": null, "developer": [ "google" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "MediaPipe legacy models for shoes, chairs, cameras and cups; support ended March 2023 per the coverage entry.", "context_length": null, "release_date": "2020-12-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "On-device 3D bounding box detection for everyday objects from MediaPipe, trained on the Objectron dataset.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2012.09988", "https://github.com/google-ai-edge/mediapipe/blob/master/docs/solutions/models.md" ], "unverified": [] }, { "id": "pointnet-family", "name": "PointNet", "level": "family", "family_id": null, "developer": [ "stanford-university" ], "model_class": "vision", "modalities_in": [ "3D" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Point-cloud networks that consume raw points for classification and segmentation (PointNet, PointNet++).", "verified_on": "2026-10-12", "sources": [ "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/pointnet2", "https://modelzoo.co/model/pointnet-pointnet2-pytorch", "https://arxiv.org/html/1612.00593" ], "unverified": [] }, { "id": "pointnet", "name": "PointNet", "level": "variant", "family_id": "pointnet-family", "developer": [ "stanford-university" ], "model_class": "vision", "modalities_in": [ "3D" ], "modalities_out": [ "3D" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2016-12-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [ "mobile or NPU" ], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Network that classifies and segments raw point sets directly.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1612.00593", "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/pointnet2", "https://aihub.qualcomm.com/models/pointnet", "https://arxiv.org/html/1612.00593" ], "unverified": [] }, { "id": "pointnet-plus-plus", "name": "PointNet++", "level": "variant", "family_id": "pointnet-family", "developer": [ "stanford-university" ], "model_class": "vision", "modalities_in": [ "3D" ], "modalities_out": [ "3D" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2017-06-07", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Hierarchical PointNet that learns local features at multiple scales; a standard point-cloud backbone.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1706.02413", "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/pointnet2", "https://modelzoo.co/model/pointnet-pointnet2-pytorch", "https://arxiv.org/html/1706.02413" ], "unverified": [] }, { "id": "dgcnn", "name": "DGCNN", "level": "model", "family_id": null, "developer": [ "massachusetts-institute-of-technology" ], "model_class": "vision", "modalities_in": [ "3D" ], "modalities_out": [ "3D" ], "architecture": "GNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2018-01-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Dynamic graph CNN that builds neighbourhood graphs on point clouds for classification and segmentation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1801.07829", "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/dgcnn", "https://arxiv.org/html/1801.07829" ], "unverified": [] }, { "id": "minkunet", "name": "MinkUNet", "level": "model", "family_id": null, "developer": [ "stanford-university" ], "model_class": "vision", "modalities_in": [ "3D" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2019-04-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Sparse 4D convolutional U-Net built on the Minkowski Engine for LiDAR segmentation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1904.08755", "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/minkunet", "https://arxiv.org/html/1904.08755" ], "unverified": [] }, { "id": "cylinder3d", "name": "Cylinder3D", "level": "model", "family_id": null, "developer": [ "cuhk" ], "model_class": "vision", "modalities_in": [ "3D" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2020-11-19", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "LiDAR semantic segmentation with cylindrical partitioning and asymmetric 3D convolutions.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2011.10033", "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/cylinder3d", "https://arxiv.org/html/2011.10033" ], "unverified": [] }, { "id": "spvcnn", "name": "SPVCNN", "level": "model", "family_id": null, "developer": [ "tsinghua-university", "massachusetts-institute-of-technology" ], "model_class": "vision", "modalities_in": [ "3D" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2020-07-31", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Sparse point-voxel convolution network found by architecture search for efficient LiDAR segmentation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2007.16100", "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/spvcnn", "https://arxiv.org/html/2007.16100" ], "unverified": [] }, { "id": "paconv", "name": "PAConv", "level": "model", "family_id": null, "developer": [ "hku" ], "model_class": "vision", "modalities_in": [ "3D" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2021-03-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Position adaptive convolution that assembles dynamic kernels for point-cloud segmentation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.14635", "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/paconv", "https://arxiv.org/html/2103.14635" ], "unverified": [] }, { "id": "pointpillars", "name": "PointPillars", "level": "model", "family_id": null, "developer": [ "nutonomy" ], "model_class": "vision", "modalities_in": [ "3D" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2018-12-14", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fast LiDAR 3D object detector that encodes points into vertical pillars; widely used in driving stacks.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1812.05784", "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/pointpillars", "https://github.com/tensorflow/models/tree/master/official/projects/pointpillars", "https://arxiv.org/html/1812.05784" ], "unverified": [] }, { "id": "ssn-shape-signature", "name": "SSN (Shape Signature Networks)", "level": "model", "family_id": null, "developer": [ "cuhk" ], "model_class": "vision", "modalities_in": [ "3D" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2020-04-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multi-class LiDAR object detection using shape signatures.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2004.02774", "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/ssn", "https://arxiv.org/html/2004.02774" ], "unverified": [] }, { "id": "3dssd", "name": "3DSSD", "level": "model", "family_id": null, "developer": [ "cuhk" ], "model_class": "vision", "modalities_in": [ "3D" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2020-02-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Point-based single-stage 3D object detector without upsampling layers.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2002.10187", "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/3dssd", "https://arxiv.org/html/2002.10187" ], "unverified": [] }, { "id": "pointrcnn", "name": "PointRCNN", "level": "model", "family_id": null, "developer": [ "cuhk" ], "model_class": "vision", "modalities_in": [ "3D" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2018-12-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Two-stage 3D object detector that generates proposals directly from raw point clouds.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1812.04244", "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/point_rcnn", "https://arxiv.org/html/1812.04244" ], "unverified": [] }, { "id": "part-a2", "name": "Part-A2", "level": "model", "family_id": null, "developer": [ "cuhk" ], "model_class": "vision", "modalities_in": [ "3D" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2019-07-08", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Part-aware and part-aggregation network for LiDAR 3D object detection.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1907.03670", "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/parta2", "https://arxiv.org/html/1907.03670" ], "unverified": [] }, { "id": "centerpoint", "name": "CenterPoint", "level": "model", "family_id": null, "developer": [ "ut-austin" ], "model_class": "vision", "modalities_in": [ "3D" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2020-06-19", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Center-based 3D object detection and tracking from LiDAR; a common production baseline.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2006.11275", "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/centerpoint", "https://arxiv.org/html/2006.11275" ], "unverified": [] }, { "id": "centerformer", "name": "CenterFormer", "level": "model", "family_id": null, "developer": [ "ucf", "tusimple" ], "model_class": "vision", "modalities_in": [ "3D" ], "modalities_out": [ "3D" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2022-09-12", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Center-based transformer for LiDAR 3D object detection.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2209.05588", "https://github.com/open-mmlab/mmdetection3d/tree/main/projects/CenterFormer", "https://arxiv.org/html/2209.05588" ], "unverified": [] }, { "id": "pv-rcnn-family", "name": "PV-RCNN", "level": "family", "family_id": null, "developer": [ "cuhk", "sensetime" ], "model_class": "vision", "modalities_in": [ "3D" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Point-voxel feature set abstraction detectors for LiDAR (PV-RCNN, PV-RCNN++).", "verified_on": "2026-10-12", "sources": [ "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/pv_rcnn", "https://github.com/open-mmlab/OpenPCDet/tree/master/tools/cfgs/waymo_models/pv_rcnn_plusplus.yaml", "https://arxiv.org/html/1912.13192" ], "unverified": [] }, { "id": "pv-rcnn", "name": "PV-RCNN", "level": "variant", "family_id": "pv-rcnn-family", "developer": [ "cuhk", "sensetime" ], "model_class": "vision", "modalities_in": [ "3D" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2019-12-31", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "LiDAR 3D detector combining voxel CNN and PointNet-style set abstraction.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1912.13192", "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/pv_rcnn", "https://arxiv.org/html/1912.13192" ], "unverified": [] }, { "id": "pv-rcnn-plus-plus", "name": "PV-RCNN++", "level": "variant", "family_id": "pv-rcnn-family", "developer": [ "cuhk", "max-planck-institute-for-informatics" ], "model_class": "vision", "modalities_in": [ "3D" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenPCDet (Apache-2.0 per README).", "context_length": null, "release_date": "2021-01-31", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Faster, more accurate PV-RCNN with local vector representation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2102.00463", "https://github.com/open-mmlab/OpenPCDet/tree/master/tools/cfgs/waymo_models/pv_rcnn_plusplus.yaml", "https://arxiv.org/html/2102.00463" ], "unverified": [] }, { "id": "votenet-family", "name": "VoteNet", "level": "family", "family_id": null, "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "3D" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Indoor 3D detection by deep Hough voting (VoteNet, ImVoteNet).", "verified_on": "2026-10-12", "sources": [ "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/votenet", "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/imvotenet", "https://arxiv.org/html/1904.09664" ], "unverified": [] }, { "id": "votenet", "name": "VoteNet", "level": "variant", "family_id": "votenet-family", "developer": [ "meta" ], "model_class": "vision", "modalities_in": [ "3D" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2019-04-21", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Deep Hough voting detector for 3D objects in indoor point clouds.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1904.09664", "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/votenet", "https://arxiv.org/html/1904.09664" ], "unverified": [] }, { "id": "imvotenet", "name": "ImVoteNet", "level": "variant", "family_id": "votenet-family", "developer": [ "meta", "stanford-university" ], "model_class": "vision", "modalities_in": [ "3D", "image" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2020-01-29", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "VoteNet boosted with image votes for indoor 3D detection.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2001.10692", "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/imvotenet", "https://arxiv.org/html/2001.10692" ], "unverified": [] }, { "id": "h3dnet", "name": "H3DNet", "level": "model", "family_id": null, "developer": [ "ut-austin" ], "model_class": "vision", "modalities_in": [ "3D" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2020-06-10", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Indoor 3D object detection using hybrid geometric primitives.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2006.05682", "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/h3dnet", "https://arxiv.org/html/2006.05682" ], "unverified": [] }, { "id": "group-free-3d", "name": "Group-Free-3D", "level": "model", "family_id": null, "developer": [ "university-of-science-and-technology-of-china", "microsoft-research" ], "model_class": "vision", "modalities_in": [ "3D" ], "modalities_out": [ "3D" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2021-04-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Transformer-based indoor 3D object detection without hand-crafted grouping.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2104.00678", "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/groupfree3d", "https://arxiv.org/html/2104.00678" ], "unverified": [] }, { "id": "fcaf3d", "name": "FCAF3D", "level": "model", "family_id": null, "developer": [ "samsung" ], "model_class": "vision", "modalities_in": [ "3D" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2021-12-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fully convolutional anchor-free indoor 3D object detection.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2112.00322", "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/fcaf3d", "https://arxiv.org/html/2112.00322" ], "unverified": [] }, { "id": "tr3d", "name": "TR3D", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "3D" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2023-02-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Real-time indoor 3D object detection.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2302.02858", "https://github.com/open-mmlab/mmdetection3d/tree/main/projects/TR3D" ], "unverified": [] }, { "id": "imvoxelnet", "name": "ImVoxelNet", "level": "model", "family_id": null, "developer": [ "samsung" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2021-06-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Camera-only 3D detection that projects image features into voxels, for monocular and multi-view input.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2106.01178", "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/imvoxelnet", "https://arxiv.org/html/2106.01178" ], "unverified": [] }, { "id": "smoke-mono3d", "name": "SMOKE", "level": "model", "family_id": null, "developer": [ "zongmu-technology" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2020-02-24", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Single-stage monocular 3D object detection via keypoint estimation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2002.10111", "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/smoke", "https://arxiv.org/html/2002.10111" ], "unverified": [] }, { "id": "fcos3d", "name": "FCOS3D", "level": "model", "family_id": null, "developer": [ "cuhk" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2021-04-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fully convolutional one-stage monocular 3D object detector.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2104.10956", "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/fcos3d", "https://arxiv.org/html/2104.10956" ], "unverified": [] }, { "id": "pgd-mono3d", "name": "PGD (Probabilistic and Geometric Depth)", "level": "model", "family_id": null, "developer": [ "cuhk" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2021-07-29", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Monocular 3D detector with probabilistic and geometric depth estimation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2107.14160", "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/pgd", "https://arxiv.org/html/2107.14160" ], "unverified": [] }, { "id": "monoflex", "name": "MonoFlex", "level": "model", "family_id": null, "developer": [ "tsinghua-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2021-04-06", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Flexible monocular 3D object detector that treats truncated objects separately.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2104.02323", "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/monoflex", "https://arxiv.org/html/2104.02323" ], "unverified": [] }, { "id": "detr3d", "name": "DETR3D", "level": "model", "family_id": null, "developer": [ "massachusetts-institute-of-technology", "toyota-research-institute" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2021-10-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multi-camera 3D object detection with 3D-to-2D queries.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2110.06922", "https://github.com/open-mmlab/mmdetection3d/tree/main/projects/DETR3D", "https://arxiv.org/html/2110.06922" ], "unverified": [] }, { "id": "mvx-net", "name": "MVX-Net", "level": "model", "family_id": null, "developer": [ "jhu", "apple" ], "model_class": "vision", "modalities_in": [ "3D", "image" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2019-04-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multimodal VoxelNet that fuses LiDAR and camera features for 3D detection.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1904.01649", "https://github.com/open-mmlab/mmdetection3d/tree/main/configs/mvxnet", "https://arxiv.org/html/1904.01649" ], "unverified": [] }, { "id": "bevfusion", "name": "BEVFusion", "level": "model", "family_id": null, "developer": [ "massachusetts-institute-of-technology" ], "model_class": "vision", "modalities_in": [ "3D", "image" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2022-05-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Multi-task multi-sensor fusion in a unified bird's-eye-view representation; a standard LiDAR-camera fusion baseline.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2205.13542", "https://github.com/open-mmlab/mmdetection3d/tree/main/projects/BEVFusion", "https://arxiv.org/html/2205.13542" ], "unverified": [] }, { "id": "tpvformer", "name": "TPVFormer", "level": "model", "family_id": null, "developer": [ "tsinghua-university", "phigent-robotics" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenMMLab MMDetection3D; repository licence Apache-2.0, weight terms not separately stated.", "context_length": null, "release_date": "2023-02-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Camera-only 3D semantic occupancy prediction with a tri-perspective view representation.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2302.07817", "https://github.com/open-mmlab/mmdetection3d/tree/main/projects/TPVFormer", "https://arxiv.org/html/2302.07817" ], "unverified": [] }, { "id": "voxel-rcnn", "name": "Voxel R-CNN", "level": "model", "family_id": null, "developer": [], "model_class": "vision", "modalities_in": [ "3D" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenPCDet (Apache-2.0 per README).", "context_length": null, "release_date": "2020-12-31", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "High-performance voxel-based two-stage LiDAR 3D detector.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2012.15712", "https://github.com/open-mmlab/OpenPCDet/tree/master/tools/cfgs/kitti_models/voxel_rcnn_car.yaml" ], "unverified": [] }, { "id": "focals-conv", "name": "Focals Conv", "level": "model", "family_id": null, "developer": [ "cuhk", "megvii" ], "model_class": "vision", "modalities_in": [ "3D" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenPCDet (Apache-2.0 per README).", "context_length": null, "release_date": "2022-04-26", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Focal sparse convolution for 3D object detection.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2204.12463", "https://github.com/open-mmlab/OpenPCDet/tree/master/tools/cfgs/kitti_models/voxel_rcnn_car_focal_multimodal.yaml", "https://arxiv.org/html/2204.12463" ], "unverified": [] }, { "id": "caddn", "name": "CaDDN", "level": "model", "family_id": null, "developer": [ "university-of-toronto" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenPCDet (Apache-2.0 per README).", "context_length": null, "release_date": "2021-03-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Monocular 3D detection with categorical depth distributions.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2103.01100", "https://github.com/open-mmlab/OpenPCDet/tree/master/tools/cfgs/kitti_models/CaDDN.yaml", "https://arxiv.org/html/2103.01100" ], "unverified": [] }, { "id": "voxelnext", "name": "VoxelNeXt", "level": "model", "family_id": null, "developer": [ "cuhk", "megvii" ], "model_class": "vision", "modalities_in": [ "3D" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenPCDet (Apache-2.0 per README).", "context_length": null, "release_date": "2023-03-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Fully sparse voxel network for LiDAR 3D detection and tracking.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2303.11301", "https://github.com/open-mmlab/OpenPCDet/tree/master/tools/cfgs/waymo_models/voxelnext_ioubranch_large.yaml", "https://arxiv.org/html/2303.11301" ], "unverified": [] }, { "id": "dsvt", "name": "DSVT", "level": "model", "family_id": null, "developer": [ "peking-university" ], "model_class": "vision", "modalities_in": [ "3D" ], "modalities_out": [ "3D" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenPCDet (Apache-2.0 per README).", "context_length": null, "release_date": "2023-01-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Dynamic sparse voxel transformer designed to be deployment friendly for LiDAR detection.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2301.06051", "https://github.com/open-mmlab/OpenPCDet/tree/master/tools/cfgs/waymo_models/dsvt_voxel.yaml", "https://arxiv.org/html/2301.06051" ], "unverified": [] }, { "id": "transfusion", "name": "TransFusion", "level": "model", "family_id": null, "developer": [ "hkust" ], "model_class": "vision", "modalities_in": [ "3D", "image" ], "modalities_out": [ "3D" ], "architecture": "dense transformer", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Checkpoints in OpenPCDet (Apache-2.0 per README).", "context_length": null, "release_date": "2022-03-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Transformer-based LiDAR-camera fusion for 3D object detection; TransFusion-L is the LiDAR-only version.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2203.11496", "https://github.com/open-mmlab/OpenPCDet/tree/master/tools/cfgs/nuscenes_models/transfusion_lidar.yaml", "https://arxiv.org/html/2203.11496" ], "unverified": [] }, { "id": "smpl-x", "name": "SMPL-X and SMPLify-X", "level": "model", "family_id": null, "developer": [ "mpi-is" ], "model_class": "3D", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Listed in OpenMMLab MMHuman3D (Apache-2.0; some methods carry additional licences).", "context_length": null, "release_date": "2019-04-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Expressive parametric body model with hands and face, and the SMPLify-X method that fits it to a single image.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1904.05866", "https://github.com/open-mmlab/mmhuman3d/tree/main/", "https://arxiv.org/html/1904.05866" ], "unverified": [] }, { "id": "mano-hand-model", "name": "MANO", "level": "model", "family_id": null, "developer": [ "mpi-is" ], "model_class": "3D", "modalities_in": [], "modalities_out": [ "3D" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Listed in OpenMMLab MMHuman3D (Apache-2.0; some methods carry additional licences).", "context_length": null, "release_date": "2017-11-22", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Parametric 3D hand model learned from scans, used with body models for hand capture.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2201.02610", "https://github.com/open-mmlab/mmhuman3d/tree/main/", "https://arxiv.org/html/2201.02610", "https://mano.is.tue.mpg.de/" ], "unverified": [] }, { "id": "star-body-model", "name": "STAR", "level": "model", "family_id": null, "developer": [ "mpi-is" ], "model_class": "3D", "modalities_in": [], "modalities_out": [ "3D" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Listed in OpenMMLab MMHuman3D (Apache-2.0; some methods carry additional licences).", "context_length": null, "release_date": "2020-08-19", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Sparse trained articulated human body regressor, a compact parametric body model.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2008.08535", "https://github.com/open-mmlab/mmhuman3d/tree/main/", "https://arxiv.org/html/2008.08535" ], "unverified": [] }, { "id": "smplify", "name": "SMPLify", "level": "model", "family_id": null, "developer": [ "microsoft-research", "mpi-is" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": "other", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Listed in OpenMMLab MMHuman3D (Apache-2.0; some methods carry additional licences).", "context_length": null, "release_date": "2016-07-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Automatic fitting of the SMPL body model to 2D joints to estimate 3D pose and shape from one image.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1607.08128", "https://github.com/open-mmlab/mmhuman3d/tree/main/", "https://arxiv.org/html/1607.08128" ], "unverified": [] }, { "id": "hmr", "name": "HMR (Human Mesh Recovery)", "level": "model", "family_id": null, "developer": [ "uc-berkeley" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Listed in OpenMMLab MMHuman3D (Apache-2.0; some methods carry additional licences).", "context_length": null, "release_date": "2017-12-18", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "End-to-end recovery of 3D human body shape and pose from a single image.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1712.06584", "https://github.com/open-mmlab/mmhuman3d/tree/main/", "https://arxiv.org/html/1712.06584" ], "unverified": [] }, { "id": "spin", "name": "SPIN", "level": "model", "family_id": null, "developer": [ "upenn" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Listed in OpenMMLab MMHuman3D (Apache-2.0; some methods carry additional licences).", "context_length": null, "release_date": "2019-09-27", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Human pose and shape regression trained with model fitting in the loop.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1909.12828", "https://github.com/open-mmlab/mmhuman3d/tree/main/", "https://arxiv.org/html/1909.12828" ], "unverified": [] }, { "id": "vibe", "name": "VIBE", "level": "model", "family_id": null, "developer": [ "mpi-is" ], "model_class": "vision", "modalities_in": [ "video" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Listed in OpenMMLab MMHuman3D (Apache-2.0; some methods carry additional licences).", "context_length": null, "release_date": "2019-12-11", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Video inference of human body pose and shape with an adversarial motion prior.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/1912.05656", "https://github.com/open-mmlab/mmhuman3d/tree/main/", "https://arxiv.org/html/1912.05656" ], "unverified": [] }, { "id": "hybrik", "name": "HybrIK", "level": "model", "family_id": null, "developer": [ "shanghai-jiao-tong-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Listed in OpenMMLab MMHuman3D (Apache-2.0; some methods carry additional licences).", "context_length": null, "release_date": "2020-11-30", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Hybrid analytical and neural inverse kinematics for 3D human pose and shape.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2011.14672", "https://github.com/open-mmlab/mmhuman3d/tree/main/", "https://arxiv.org/html/2011.14672" ], "unverified": [] }, { "id": "pare", "name": "PARE", "level": "model", "family_id": null, "developer": [ "mpi-is", "eth-zurich" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Listed in OpenMMLab MMHuman3D (Apache-2.0; some methods carry additional licences).", "context_length": null, "release_date": "2021-04-17", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Part attention regressor for robust 3D human body estimation under occlusion.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2104.08527", "https://github.com/open-mmlab/mmhuman3d/tree/main/", "https://arxiv.org/html/2104.08527" ], "unverified": [] }, { "id": "expose", "name": "ExPose", "level": "model", "family_id": null, "developer": [ "mpi-is" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Listed in OpenMMLab MMHuman3D (Apache-2.0; some methods carry additional licences).", "context_length": null, "release_date": "2020-08-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Monocular regression of an expressive body with hands and face using body-driven attention.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2008.09062", "https://github.com/open-mmlab/mmhuman3d/tree/main/configs/expose", "https://arxiv.org/html/2008.09062" ], "unverified": [] }, { "id": "pymaf-x", "name": "PyMAF-X", "level": "model", "family_id": null, "developer": [ "tsinghua-university" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Listed in OpenMMLab MMHuman3D (Apache-2.0; some methods carry additional licences).", "context_length": null, "release_date": "2022-07-13", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Full-body model regression from monocular images with well-aligned body, hands and face.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2207.06400", "https://github.com/open-mmlab/mmhuman3d/tree/main/configs/pymafx", "https://arxiv.org/html/2207.06400" ], "unverified": [] }, { "id": "cliff", "name": "CLIFF", "level": "model", "family_id": null, "developer": [ "huawei-technologies" ], "model_class": "vision", "modalities_in": [ "image" ], "modalities_out": [ "3D" ], "architecture": "CNN", "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Listed in OpenMMLab MMHuman3D (Apache-2.0; some methods carry additional licences).", "context_length": null, "release_date": "2022-08-01", "release_date_basis": "announcement page", "weights_date": null, "openness": "unknown", "licence": null, "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Human pose and shape estimation that uses location information in the full frame.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2208.00571", "https://github.com/open-mmlab/mmhuman3d/tree/main/configs/cliff", "https://arxiv.org/html/2208.00571" ], "unverified": [] }, { "id": "gigaai-giga-world-1", "name": "GigaWorld-1", "level": "model", "family_id": null, "developer": [ "gigaai", "tsinghua-university" ], "model_class": "world model", "modalities_in": [ "image", "text", "action" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2026-07-01", "release_date_basis": "registry date", "weights_date": "2026-07-01", "openness": "open weights permissive", "licence": "apache-2-0", "access_routes": [ "open weights download" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "World model released by GigaAI for evaluating robot policies in simulated video.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2607.02642", "https://huggingface.co/open-gigaai/Giga-World-1", "https://huggingface.co/api/models?pipeline_tag=robotics&sort=likes&limit=50", "https://arxiv.org/html/2607.02642" ], "unverified": [] }, { "id": "bria-ai-bria-3-2", "name": "Bria 3.2", "level": "model", "family_id": "bria-ai-bria", "developer": [ "bria-ai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": 4000000000, "params_active": null, "params_checkpoint": null, "params_note": "4B parameters (Replicate page).", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "open weights restricted", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Commercial text-to-image model trained on licensed data, with 4B parameters and a focus on aesthetics and text rendering.", "verified_on": "2026-10-12", "sources": [ "https://huggingface.co/briaai/BRIA-3.2", "https://replicate.com/bria/image-3.2" ], "unverified": [] }, { "id": "openai-dall-e-3", "name": "DALL-E 3", "level": "model", "family_id": "openai-dall-e", "developer": [ "openai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2023-09", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "OpenAI system that creates realistic images and art from a natural-language description.", "verified_on": "2026-10-12", "sources": [ "https://openai.com/index/dall-e-3/", "https://platform.openai.com/docs/models/dall-e-3", "https://replicate.com/openai/dall-e-3" ], "unverified": [ "release_date" ] }, { "id": "openai-gpt-image-1", "name": "GPT Image 1", "level": "model", "family_id": "openai-gpt-image", "developer": [ "openai" ], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "OpenAI multimodal image generation model; on Replicate it runs with the user's own OpenAI key.", "verified_on": "2026-10-12", "sources": [ "https://developers.openai.com/api/docs/models/gpt-image-1", "https://openai.com/index/image-generation-api/", "https://replicate.com/openai/gpt-image-1" ], "unverified": [] }, { "id": "openai-gpt-image-1-mini", "name": "GPT Image 1 mini", "level": "model", "family_id": "openai-gpt-image", "developer": [ "openai" ], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "deprecated", "summary": "Cost-efficient version of OpenAI GPT Image 1.", "verified_on": "2026-10-12", "sources": [ "https://platform.openai.com/docs/models/gpt-image-1-mini", "https://replicate.com/openai/gpt-image-1-mini" ], "unverified": [] }, { "id": "minimax-image-01", "name": "MiniMax Image-01", "level": "model", "family_id": null, "developer": [ "minimax" ], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-02-15", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "MiniMax first image model, with character reference support.", "verified_on": "2026-10-12", "sources": [ "https://www.minimax.io/news/image-01", "https://platform.minimax.io/docs/release-notes/models", "https://replicate.com/minimax/image-01" ], "unverified": [] }, { "id": "recraft-20b", "name": "Recraft 20B", "level": "model", "family_id": "recraft", "developer": [ "recraft" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2024-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Lower-cost, faster Recraft image model.", "verified_on": "2026-10-12", "sources": [ "https://webflow.recraft.ai/docs", "https://replicate.com/recraft-ai/recraft-20b" ], "unverified": [] }, { "id": "ideogram-4-5", "name": "Ideogram 4.5", "level": "model", "family_id": "ideogram", "developer": [ "ideogram" ], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Replicate also lists Ideogram Character and Layerize.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Ideogram image generation and editing model, newer than Ideogram 4.0.", "verified_on": "2026-10-12", "sources": [ "https://ideogram.ai/models/4.5/", "https://replicate.com/ideogram-ai/ideogram-4-5" ], "unverified": [] }, { "id": "leonardo-ai-phoenix-1-0", "name": "Leonardo Phoenix 1.0", "level": "model", "family_id": null, "developer": [ "leonardo-ai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Leonardo.Ai first foundational image model, producing images up to 5 megapixels in fast, quality and ultra modes.", "verified_on": "2026-10-12", "sources": [ "https://leonardo.ai/phoenix", "https://docs.leonardo.ai/docs/phoenix", "https://replicate.com/leonardoai/phoenix-1.0" ], "unverified": [] }, { "id": "leonardo-ai-lucid-origin", "name": "Leonardo Lucid Origin", "level": "model", "family_id": null, "developer": [ "leonardo-ai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Leonardo.Ai image model aimed at artistic, high-quality visuals with improved prompt adherence and diversity.", "verified_on": "2026-10-12", "sources": [ "https://leonardo.ai/news/lucid-origin-ai-image-model", "https://replicate.com/leonardoai/lucid-origin" ], "unverified": [] }, { "id": "leonardo-ai-motion-2-0", "name": "Leonardo Motion 2.0", "level": "model", "family_id": null, "developer": [ "leonardo-ai" ], "model_class": "video generation", "modalities_in": [ "text" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Leonardo.Ai text-to-video model that creates 5-second 480p clips.", "verified_on": "2026-10-12", "sources": [ "https://docs.leonardo.ai/docs/motion-20", "https://replicate.com/leonardoai/motion-2.0" ], "unverified": [] }, { "id": "quiverai-arrow-1-1", "name": "Arrow 1.1", "level": "model", "family_id": null, "developer": [ "quiverai" ], "model_class": "image generation", "modalities_in": [ "text" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "QuiverAI model that generates editable SVG vector graphics such as icons, logos and illustrations from text prompts.", "verified_on": "2026-10-12", "sources": [ "https://quiver.ai/blog/introducing-arrow-1-1/", "https://replicate.com/quiverai/arrow-1.1" ], "unverified": [] }, { "id": "sourceful-riverflow-2-0", "name": "Riverflow 2.0", "level": "model", "family_id": null, "developer": [ "sourceful" ], "model_class": "image generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "image" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Replicate also lists Riverflow 2.5.", "context_length": null, "release_date": "2026-02-02", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Agentic image generation and editing model from Sourceful built for precise brand work with font control.", "verified_on": "2026-10-12", "sources": [ "https://www.sourceful.com/research", "https://replicate.com/sourceful/riverflow-2.0-pro" ], "unverified": [] }, { "id": "decart-lucy-edit-2", "name": "Lucy Edit 2", "level": "model", "family_id": null, "developer": [ "decart" ], "model_class": "video generation", "modalities_in": [ "video", "text", "image" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Decart video editing model for style transfer, object replacement and character changes from text prompts and reference images.", "verified_on": "2026-10-12", "sources": [ "https://docs.platform.decart.ai/changelog", "https://decart.ai/lucy", "https://replicate.com/decart/lucy-edit-2" ], "unverified": [] }, { "id": "kuaishou-kling-2-5-turbo", "name": "Kling 2.5 Turbo", "level": "model", "family_id": "kuaishou-kling", "developer": [ "kuaishou" ], "model_class": "video generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-09-23", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Kuaishou text-to-video and image-to-video model aimed at smooth motion and prompt adherence.", "verified_on": "2026-10-12", "sources": [ "https://ir.kuaishou.com/news-releases/news-release-details/kling-ai-launches-25-turbo-video-model-industry-leading", "https://replicate.com/kwaivgi/kling-v2.5-turbo-pro" ], "unverified": [ "release_date" ] }, { "id": "kuaishou-kling-2-6", "name": "Kling 2.6", "level": "model", "family_id": "kuaishou-kling", "developer": [ "kuaishou" ], "model_class": "video generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "video", "audio" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": "2025-12-03", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Kuaishou video model that generates video together with native audio (speech, sound effects and ambience) from text or an image.", "verified_on": "2026-10-12", "sources": [ "https://ir.kuaishou.com/news-releases/news-release-details/kling-ai-launches-video-26-model-simultaneous-audio-visual", "https://app.klingai.com/global/release-notes/c605hp1tzd?type=dialog", "https://replicate.com/kwaivgi/kling-v2.6" ], "unverified": [ "release_date" ] }, { "id": "luma-ai-ray-flash-2", "name": "Ray Flash 2", "level": "model", "family_id": null, "developer": [ "luma-ai" ], "model_class": "video generation", "modalities_in": [ "text", "image" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Faster, cheaper version of Luma Ray2 for 5 s and 9 s 720p video.", "verified_on": "2026-10-12", "sources": [ "https://lumalabs.ai/learning-hub/dream-machine-guide-ray2", "https://docs.lumalabs.ai/docs/video-generation", "https://replicate.com/luma/ray-flash-2-720p" ], "unverified": [] }, { "id": "pruna-ai-p-video", "name": "P-Video", "level": "model", "family_id": null, "developer": [ "pruna-ai" ], "model_class": "video generation", "modalities_in": [ "text", "image", "audio" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Replicate also lists P-Video 2.", "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Pruna AI fast video generation model with a draft mode, covering text, image and audio to video.", "verified_on": "2026-10-12", "sources": [ "https://www.pruna.ai/p-video", "https://www.pruna.ai/blog/p-video-fast-cheap-draft", "https://replicate.com/prunaai/p-video" ], "unverified": [] }, { "id": "bytedance-dreamactor-m2-0", "name": "DreamActor-M2.0", "level": "model", "family_id": "bytedance-dreamactor", "developer": [ "bytedance" ], "model_class": "avatar", "modalities_in": [ "image", "video" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Animates any character, including humans, cartoons and animals, from a single image and a driving video.", "verified_on": "2026-10-12", "sources": [ "https://arxiv.org/abs/2601.21716", "https://replicate.com/bytedance/dreamactor-m2.0" ], "unverified": [] }, { "id": "sync-labs-lipsync-2", "name": "lipsync-2", "level": "model", "family_id": null, "developer": [ "sync-labs" ], "model_class": "avatar", "modalities_in": [ "video", "audio" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": "Replicate lists lipsync-2-pro.", "context_length": null, "release_date": "2025-08-20", "release_date_basis": "announcement page", "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Sync Labs lip-sync model that matches mouth movement in a video to new audio, for dubbing.", "verified_on": "2026-10-12", "sources": [ "https://sync.so/blog/lipsync-2-0", "https://sync.so/docs/models/lipsync", "https://replicate.com/sync/lipsync-2-pro" ], "unverified": [] }, { "id": "sync-labs-react-1", "name": "react-1", "level": "model", "family_id": null, "developer": [ "sync-labs" ], "model_class": "avatar", "modalities_in": [ "video", "audio" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Sync Labs lip-sync model with refined control of human emotion.", "verified_on": "2026-10-12", "sources": [ "https://sync.so/react-1", "https://sync.so/docs/models/react", "https://replicate.com/sync/react-1" ], "unverified": [] }, { "id": "veed-fabric-1-0", "name": "VEED Fabric 1.0", "level": "model", "family_id": null, "developer": [ "veed" ], "model_class": "avatar", "modalities_in": [ "image", "audio" ], "modalities_out": [ "video" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": "proprietary", "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "VEED image-to-video API that turns any image into a talking video.", "verified_on": "2026-10-12", "sources": [ "https://www.veed.io/learn/fabric-1-0-api", "https://replicate.com/veed/fabric-1.0" ], "unverified": [] }, { "id": "tencent-hunyuan3d-3-1", "name": "Hunyuan3D 3.1", "level": "variant", "family_id": "tencent-hunyuan3d", "developer": [ "tencent" ], "model_class": "3D", "modalities_in": [ "image", "text" ], "modalities_out": [ "3D" ], "architecture": null, "architecture_detail": null, "params_total": null, "params_active": null, "params_checkpoint": null, "params_note": null, "context_length": null, "release_date": null, "release_date_basis": null, "weights_date": null, "openness": "proprietary", "licence": null, "access_routes": [ "API" ], "deployment_tiers": [], "footprint_note": null, "runtimes": [], "languages_note": null, "indonesian_support": "not stated", "regional_languages": [], "indonesia_hosting": "unknown", "status": "active", "summary": "Tencent 3D generation model aimed at texture fidelity and geometry precision.", "verified_on": "2026-10-15", "sources": [ "https://www.tencentcloud.com/document/product/1284/75540", "https://replicate.com/tencent/hunyuan-3d-3.1" ], "unverified": [] } ]