[ { "id": "aa-briefcase-v1-1", "name": "AA-Briefcase", "version": "v1.1", "category": "composite index", "measures": "Component eval of the AA Intelligence Index v4.3.2", "maintainer": "artificial-analysis", "status": "active", "url": null, "sources": [ "https://artificialanalysis.ai/" ], "unverified": [] }, { "id": "aa-image-arena", "name": "Artificial Analysis Image Arena", "version": null, "category": "generative media", "measures": "Human preference for image generation", "maintainer": "artificial-analysis", "status": "active", "url": null, "sources": [ "https://artificialanalysis.ai/text-to-video/arena" ], "unverified": [] }, { "id": "aa-image-t2i-v2-0", "name": "Artificial Analysis Text-to-Image Arena", "version": "AA-Image-T2I v2.0", "category": "generative media", "measures": "Elo from blind pairwise preference votes on text-to-image outputs", "maintainer": "artificial-analysis", "status": "active", "url": "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text", "sources": [ "https://artificialanalysis.ai/text-to-image/arena/leaderboard-text" ] }, { "id": "aa-intelligence-index", "name": "Artificial Analysis Intelligence Index", "version": "v4.3.2", "category": "composite index", "measures": "Composite intelligence score across AA evaluations", "maintainer": "artificial-analysis", "status": "active", "url": "https://artificialanalysis.ai/leaderboards/models", "sources": [ "https://artificialanalysis.ai/leaderboards/models", "https://artificialanalysis.ai/models/claude-opus-5-5", "https://slash-digital.io/en/insights/ai-models-2026/", "https://artificialanalysis.ai/leaderboards/models?open_weights=true", "https://artificialanalysis.ai/models/open-source", "https://artificialanalysis.ai/", "https://www.datalearner.com/en/leaderboards/external/aa-quality-index" ] }, { "id": "aa-lcr-v1-1", "name": "AA-LCR v1.1", "version": "v1.1", "category": "long context", "measures": "Long-context reasoning; part of AA Intelligence Index", "maintainer": "artificial-analysis", "status": "active", "url": "https://artificialanalysis.ai/", "sources": [ "https://artificialanalysis.ai/" ], "unverified": [] }, { "id": "aa-music-arena-v1-1", "name": "Artificial Analysis Music Arena", "version": "v1.1", "category": "generative media", "measures": "Elo from recruited blind-panel votes (37,000+ votes, 1,000 prompts, 17 genres) on Vocal and Instrumental boards; MiniMax Music 3.0 anchored at 1000", "maintainer": "artificial-analysis", "status": "active", "url": null, "sources": [ "https://alphasignal.ai/news/artificial-analysis-music-arena-v1-1-crowns-suno-v6-across-1-000-harder-prompts", "https://digg.com/ai/g8egglmo" ] }, { "id": "aa-omniscience", "name": "AA-Omniscience", "version": null, "category": "composite index", "measures": "Component eval of the AA Intelligence Index v4.3.2", "maintainer": "artificial-analysis", "status": "active", "url": null, "sources": [ "https://artificialanalysis.ai/" ], "unverified": [] }, { "id": "aa-speech-arena", "name": "Artificial Analysis Speech Arena", "version": null, "category": "speech", "measures": "Human preference votes between text-to-speech samples", "maintainer": "artificial-analysis", "status": null, "url": "https://artificialanalysis.ai/text-to-speech/arena", "sources": [ "https://artificialanalysis.ai/text-to-speech/arena" ] }, { "id": "aa-tts", "name": "Artificial Analysis Text-to-Speech Arena", "version": null, "category": "generative media", "measures": "Elo from blind preference votes on speech outputs; 96 models, 16 open weights", "maintainer": "artificial-analysis", "status": "active", "url": "https://artificialanalysis.ai/text-to-speech/leaderboard", "sources": [ "https://artificialanalysis.ai/text-to-speech/leaderboard" ] }, { "id": "aa-video-arena", "name": "Artificial Analysis Video Arena", "version": null, "category": "generative media", "measures": "Human preference for video generation", "maintainer": "artificial-analysis", "status": "active", "url": "https://artificialanalysis.ai/text-to-video/arena", "sources": [ "https://artificialanalysis.ai/text-to-video/arena" ], "unverified": [] }, { "id": "aa-video-i2v-v1-0", "name": "Artificial Analysis Image-to-Video leaderboard", "version": "AA-Video-I2V v1.0", "category": "generative media", "measures": "Elo from preference votes on image-to-video outputs", "maintainer": "artificial-analysis", "status": "active", "url": null, "sources": [ "https://alphasignal.ai/news/shengshu-s-vidu-q4-preview-debuts-at-no-3-for-ai-video" ] }, { "id": "aa-video-t2v", "name": "Artificial Analysis Text-to-Video leaderboard", "version": null, "category": "generative media", "measures": "Elo from blind preference votes on text-to-video outputs; Kling 3.0 1080p (Pro) anchored at 1000", "maintainer": "artificial-analysis", "status": "active", "url": "https://artificialanalysis.ai/video/leaderboard/text-to-video", "sources": [ "https://artificialanalysis.ai/video/leaderboard/text-to-video" ] }, { "id": "aegis-2-0", "name": "Aegis", "version": "2.0", "category": "safety", "measures": null, "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/nvidia/Nemotron-3.5-Content-Safety" ] }, { "id": "afrimed-qa", "name": "AfriMed-QA", "version": null, "category": "domain", "measures": "African medical QA", "maintainer": null, "status": null, "url": null, "sources": [ "https://arxiv.org/html/2507.05201v2" ] }, { "id": "agentclinic-medqa", "name": "AgentClinic-MedQA", "version": null, "category": "domain", "measures": "Agentic clinical dialogue", "maintainer": null, "status": null, "url": null, "sources": [ "https://arxiv.org/html/2507.05201v2" ] }, { "id": "agentharm", "name": "AgentHarm", "version": null, "category": "safety", "measures": "Harmfulness of LLM agents: refusal of 110 malicious agent tasks (440 with augmentations) across 11 harm categories, and capability retention after jailbreak", "maintainer": null, "status": null, "url": "https://arxiv.org/abs/2410.09024", "sources": [ "https://arxiv.org/abs/2410.09024" ] }, { "id": "ai-needle", "name": "AI-Needle", "version": null, "category": "long context", "measures": "Long-context retrieval", "maintainer": null, "status": "active", "url": null, "sources": [ "https://benchlm.ai/best/long-context" ], "unverified": [] }, { "id": "ai2d", "name": "AI2D", "version": null, "category": "multimodal", "measures": null, "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/microsoft/Phi-4-reasoning-vision-15B" ] }, { "id": "ailuminate-jailbreak-v0-5", "name": "MLCommons AILuminate Jailbreak", "version": "v0.5", "category": "safety", "measures": "Jailbreak benchmark (draft)", "maintainer": "mlcommons", "status": "active", "url": "https://mlcommons.org/benchmarks/ailuminate/", "sources": [ "https://mlcommons.org/benchmarks/ailuminate/" ], "unverified": [] }, { "id": "ailuminate-safety-v1-0", "name": "MLCommons AILuminate Safety", "version": "v1.0", "category": "safety", "measures": "12 hazard categories, graded Poor to Excellent against a reference system; English and French available, Chinese and Hindi in development", "maintainer": "mlcommons", "status": "active", "url": "https://mlcommons.org/benchmarks/ailuminate/", "sources": [ "https://mlcommons.org/benchmarks/ailuminate/" ], "unverified": [] }, { "id": "aime", "name": "AIME", "version": "2024/2025/2026", "category": "math", "measures": "Answers to the American Invitational Mathematics Examination problems", "maintainer": "MAA problems; many evaluators, USA", "status": "saturated", "url": "https://www.benchlm.ai/benchmarks/aime", "sources": [ "https://www.benchlm.ai/benchmarks/aime" ], "unverified": [ "status" ] }, { "id": "aime-2024", "name": "AIME24", "version": "2024", "category": "math", "measures": "AIME 2024 problems", "maintainer": null, "status": "saturated", "url": null, "sources": [ "https://huggingface.co/microsoft/Phi-4-mini-flash-reasoning", "https://arxiv.org/html/2501.19393v1", "https://arxiv.org/abs/2502.03387", "https://benchlm.ai/blog/posts/aime-hmmt-competition-math" ], "unverified": [ "status" ] }, { "id": "aime-2025", "name": "AIME 2025", "version": "2025", "category": "math", "measures": "AIME 2025 problems", "maintainer": null, "status": "saturated", "url": null, "sources": [ "https://huggingface.co/allenai/Olmo-3-32B-Think", "https://huggingface.co/microsoft/Phi-4-mini-flash-reasoning", "https://huggingface.co/HuggingFaceTB/SmolLM3-3B" ], "unverified": [ "status" ] }, { "id": "ami", "name": "AMI (IHM, SDM)", "version": null, "category": "speech", "measures": "Speaker diarization error rate", "maintainer": null, "status": "static", "url": null, "sources": [ "https://www.pyannote.ai/blog/top-7-speaker-diarization-apis-and-models-in-2026" ] }, { "id": "androidworld", "name": "AndroidWorld", "version": null, "category": "agents and tool use", "measures": "Agent benchmark tracked by BenchLM", "maintainer": null, "status": null, "url": null, "sources": [ "https://www.benchlm.ai/benchmarks" ], "unverified": [] }, { "id": "arc-agi-2", "name": "ARC-AGI-2", "version": null, "category": "knowledge and reasoning", "measures": "Abstract reasoning puzzles", "maintainer": "ARC Prize", "status": "active", "url": "https://arcprize.org", "sources": [ "https://slash-digital.io/en/insights/ai-models-2026/" ] }, { "id": "arc-agi-3", "name": "ARC-AGI-3", "version": null, "category": "knowledge and reasoning", "measures": "Abstract reasoning puzzles; scores depend heavily on harness", "maintainer": "ARC Prize", "status": "active", "url": "https://arcprize.org", "sources": [ "https://pasqualepillitteri.it/en/news/14246/openai-launches-gpt-6-astra-pricing-benchmarks", "https://docs.apiyi.com/en/news/gpt-6-astra-launch" ] }, { "id": "arc-c", "name": "ARC-c", "version": null, "category": "knowledge and reasoning", "measures": "ARC Challenge science questions (Gemma 3n PT card)", "maintainer": null, "status": null, "url": null, "sources": [ "https://ai.google.dev/gemma/docs/gemma-3n/model_card" ], "unverified": [] }, { "id": "arena-alignment-index", "name": "Arena Alignment Index", "version": null, "category": "safety", "measures": "Alignment ranking on arena.ai", "maintainer": "lmarena", "status": "active", "url": "https://arena.ai/leaderboard", "sources": [ "https://arena.ai/leaderboard" ], "unverified": [] }, { "id": "arena-hard", "name": "Arena-Hard", "version": null, "category": "human preference arena", "measures": "Hard chat prompts judged against a baseline", "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/ai21labs/AI21-Jamba-Mini-1.5", "https://www.ai21.com/blog/announcing-jamba-model-family" ] }, { "id": "audiobench", "name": "AudioBench", "version": null, "category": "speech", "measures": "Audio LLM evaluation: ASR, speech translation, spoken QA, audio scene and paralinguistic tasks", "maintainer": "a-star", "status": "active", "url": "https://huggingface.co/spaces/MERaLiON/AudioBench-Leaderboard", "sources": [ "https://huggingface.co/MERaLiON/MERaLiON-2-10B" ] }, { "id": "automationbench", "name": "AutomationBench", "version": null, "category": "agents and tool use", "measures": null, "maintainer": null, "status": "active", "url": null, "sources": [ "https://www.orcarouter.ai/es/blog/gemini-4-argon" ] }, { "id": "automationbench-aa", "name": "AutomationBench-AA", "version": null, "category": "composite index", "measures": "Component eval of the AA Intelligence Index v4.3.2", "maintainer": "artificial-analysis", "status": "active", "url": null, "sources": [ "https://artificialanalysis.ai/" ], "unverified": [] }, { "id": "benchlm-composite", "name": "benchlm.ai frontier composite", "version": null, "category": "composite index", "measures": "Third-party composite frontier ranking", "maintainer": "benchlm.ai", "status": "active", "url": "https://benchlm.ai/frontier-ai-models", "sources": [ "https://benchlm.ai/frontier-ai-models" ] }, { "id": "benchlm-long-context", "name": "BenchLM long-context composite", "version": null, "category": "long context", "measures": "Composite of LongBench v2, MRCR v2, AI-Needle, Graphwalks, MMLongBench-Doc", "maintainer": "BenchLM.ai", "status": "active", "url": "https://benchlm.ai/best/long-context", "sources": [ "https://benchlm.ai/best/long-context" ], "unverified": [] }, { "id": "bfcl", "name": "BFCL", "version": null, "category": "agents and tool use", "measures": "Function-calling benchmark (SmolLM3 card; version not stated)", "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/HuggingFaceTB/SmolLM3-3B" ], "unverified": [] }, { "id": "bfcl-v4", "name": "BFCL v4", "version": null, "category": "agents and tool use", "measures": "Agent benchmark tracked by BenchLM", "maintainer": null, "status": null, "url": null, "sources": [ "https://www.benchlm.ai/benchmarks" ], "unverified": [] }, { "id": "browsecomp", "name": "BrowseComp", "version": null, "category": "agents and tool use", "measures": "Web browsing agent tasks", "maintainer": "openai", "status": "crowded", "url": "https://arxiv.org/abs/2504.12516", "sources": [ "https://www.minimax.io/models/text/m3", "https://huggingface.co/moonshotai/Kimi-K3", "https://arxiv.org/abs/2504.12516", "https://www.benchlm.ai/benchmarks/browsecomp" ] }, { "id": "c-eval", "name": "C-Eval", "version": null, "category": "Chinese language", "measures": "52-discipline Chinese exam questions", "maintainer": "SJTU, Tsinghua and Edinburgh authors, China", "status": "saturated", "url": "https://cevalbenchmark.com/static/leaderboard.html", "sources": [ "https://arxiv.org/abs/2305.08322", "https://cevalbenchmark.com/static/leaderboard.html" ], "unverified": [ "status", "maintainer" ] }, { "id": "c-mapss", "name": "C-MAPSS turbofan degradation", "version": null, "category": "industrial", "measures": "Remaining-useful-life prediction from simulated engine sensors; four sub-sets; no central leaderboard", "maintainer": "nasa", "status": "static", "url": "https://www.nasa.gov/intelligent-systems-division/discovery-and-systems-health/pcoe/pcoe-data-set-repository/", "sources": [ "https://www.nasa.gov/intelligent-systems-division/discovery-and-systems-health/pcoe/pcoe-data-set-repository/", "https://arxiv.org/pdf/2511.19124" ], "unverified": [] }, { "id": "calvin", "name": "CALVIN", "version": null, "category": "world models and robotics", "measures": "Long-horizon language-conditioned robot manipulation in simulation: chains of tasks from free-form instructions", "maintainer": "Mees, Hermann, Rosete, Burgard", "status": null, "url": "https://github.com/mees/calvin", "sources": [ "https://github.com/mees/calvin" ] }, { "id": "casp16", "name": "CASP16", "version": "16", "category": "domain", "measures": "Protein structure prediction assessment (monomers, complexes)", "maintainer": null, "status": "static", "url": null, "sources": [ "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12750037/", "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12750043/", "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12750027/", "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12750027/" ] }, { "id": "causalvqa", "name": "CausalVQA", "version": null, "category": "world models and robotics", "measures": "Causal and counterfactual reasoning on video", "maintainer": "meta", "status": "active", "url": "https://ai.meta.com/blog/v-jepa-2-world-model-benchmarks/", "sources": [ "https://ai.meta.com/blog/v-jepa-2-world-model-benchmarks/" ] }, { "id": "chexpert", "name": "CheXpert", "version": null, "category": "domain", "measures": "Chest X-ray classification (macro F1), used out of distribution for MedGemma", "maintainer": null, "status": null, "url": null, "sources": [ "https://arxiv.org/html/2507.05201v2" ] }, { "id": "chronos-benchmark-ii", "name": "Chronos Benchmark II", "version": null, "category": "time series", "measures": "Zero-shot time-series forecasting", "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/amazon/chronos-2" ] }, { "id": "cmmlu", "name": "CMMLU", "version": null, "category": "Chinese language", "measures": "67 subjects, including China-specific ones; contamination check for new entries since 2023-12-16", "maintainer": "Haonan Li, Timothy Baldwin et al. (MBZUAI authors), UAE/China", "status": "active", "url": "https://github.com/haonan-li/CMMLU", "sources": [ "https://github.com/haonan-li/CMMLU" ], "unverified": [] }, { "id": "coco", "name": "COCO val", "version": "val", "category": "domain", "measures": "Object detection and segmentation AP (mAP50-95)", "maintainer": null, "status": "static", "url": null, "sources": [ "https://docs.ultralytics.com/models/yolo26/", "https://arxiv.org/html/2511.09554v2" ] }, { "id": "codeforces", "name": "Codeforces rating", "version": null, "category": "coding", "measures": "Competitive programming Elo-style rating", "maintainer": "Codeforces", "status": "active", "url": null, "sources": [ "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash", "https://blog.buildfastwithai.com/deepseek-v4-pro-review-2026" ] }, { "id": "common-voice-22", "name": "Common Voice 22 (CV22-76)", "version": "22", "category": "speech", "measures": "Multilingual ASR CER", "maintainer": null, "status": "static", "url": null, "sources": [ "https://arxiv.org/pdf/2511.09690" ] }, { "id": "common-voice-6-1", "name": "Common Voice 6.1", "version": "6.1", "category": "speech", "measures": "Speech recognition word error rate on the Mozilla Common Voice test split", "maintainer": "mozilla", "status": null, "url": null, "sources": [ "https://huggingface.co/indonesian-nlp/wav2vec2-indonesian-javanese-sundanese" ] }, { "id": "common-voice-7", "name": "Common Voice 7", "version": "7.0", "category": "speech", "measures": "Speech recognition word error rate on the Mozilla Common Voice test split", "maintainer": "mozilla", "status": null, "url": null, "sources": [ "https://huggingface.co/indonesian-nlp/wav2vec2-indonesian-javanese-sundanese" ] }, { "id": "convfinqa", "name": "ConvFinQA", "version": null, "category": "domain", "measures": "Conversational financial question answering", "maintainer": null, "status": "active", "url": null, "sources": [ "https://arxiv.org/html/2305.05862v2" ] }, { "id": "critpt", "name": "CritPt", "version": null, "category": "knowledge and reasoning", "measures": "Component eval of the AA Intelligence Index v4.3.2", "maintainer": null, "status": "active", "url": null, "sources": [ "https://artificialanalysis.ai/" ], "unverified": [] }, { "id": "cursorbench", "name": "CursorBench", "version": null, "category": "coding", "measures": "Cursor vendor coding benchmark", "maintainer": "anysphere", "status": "active", "url": null, "sources": [ "https://llm-stats.com/blog/research/claude-opus-5-5-launch", "https://cursor.com/blog/composer-2" ] }, { "id": "cwru-bearing", "name": "CWRU bearing", "version": null, "category": "industrial", "measures": "Vibration fault diagnosis; segment-wise and condition-wise splits leak information, bearing-wise splits proposed (also Paderborn, Ottawa, HUST)", "maintainer": null, "status": "contaminated", "url": null, "sources": [ "https://arxiv.org/abs/2509.22267v1" ] }, { "id": "deepswe", "name": "DeepSWE (version not stated)", "version": null, "category": "coding", "measures": "Agentic software engineering", "maintainer": null, "status": "active", "url": null, "sources": [ "https://noqta.tn/en/blog/gemini-3-6-flash-antigravity-agentic-coding-2026", "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://huggingface.co/XiaomiMiMo", "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-RL" ] }, { "id": "deepswe-v1-1", "name": "DeepSWE", "version": "v1.1", "category": "coding", "measures": "Agentic software engineering", "maintainer": null, "status": "active", "url": null, "sources": [ "https://pasqualepillitteri.it/en/news/14246/openai-launches-gpt-6-astra-pricing-benchmarks", "https://www.orcarouter.ai/fr/blog/swe-2", "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash" ] }, { "id": "dihard-3", "name": "DIHARD 3", "version": null, "category": "speech", "measures": "Speaker diarization error rate (full)", "maintainer": null, "status": "static", "url": null, "sources": [ "https://www.pyannote.ai/blog/top-7-speaker-diarization-apis-and-models-in-2026" ] }, { "id": "ehrqa", "name": "EHRQA", "version": null, "category": "domain", "measures": "Question answering over electronic health records", "maintainer": null, "status": null, "url": null, "sources": [ "https://arxiv.org/abs/2604.05081" ] }, { "id": "epic-kitchens-100", "name": "Epic-Kitchens-100 action anticipation", "version": null, "category": "multimodal", "measures": "Action anticipation, recall at 5", "maintainer": null, "status": "active", "url": null, "sources": [ "https://arxiv.org/abs/2506.09985" ] }, { "id": "epoch-capabilities-index", "name": "Epoch Capabilities Index (ECI)", "version": null, "category": "composite index", "measures": "Composite that puts many benchmarks on one scale; the hub tracks 89 benchmarks", "maintainer": "epoch-ai", "status": "active", "url": "https://epoch.ai/benchmarks", "sources": [ "https://epoch.ai/benchmarks", "https://www.orcarouter.ai/de/blog/claude-sonnet-5-5-epoch-capabilities-index" ], "unverified": [] }, { "id": "fc-reward-bench", "name": "FC Reward Bench", "version": null, "category": "agents and tool use", "measures": null, "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/ibm-granite/granite-guardian-4.1-8b" ] }, { "id": "fep-plus-openfe", "name": "FEP+/OpenFE binding affinity benchmark", "version": null, "category": "domain", "measures": "Binding affinity prediction (Pearson correlation)", "maintainer": null, "status": null, "url": null, "sources": [ "https://intuitionlabs.ai/articles/open-source-ai-drug-discovery-boltz-pfizer" ] }, { "id": "fev-bench", "name": "fev-bench", "version": null, "category": "time series", "measures": "Forecasting evaluation", "maintainer": "amazon", "status": "active", "url": null, "sources": [ "https://thenewstack.io/google-timesfm-3-multivariate-forecasting/", "https://huggingface.co/amazon/chronos-2", "https://www.amazon.science/blog/introducing-chronos-2-from-univariate-to-universal-forecasting", "https://datanorth.ai/nl/nieuws/google-lanceert-timesfm-3" ] }, { "id": "fiction-livebench", "name": "Fiction.LiveBench", "version": null, "category": "long context", "measures": "Long creative-writing comprehension: theory of mind, chronology, implicit information; questions on Fiction.live stories", "maintainer": "Fiction.live", "status": null, "url": "https://fiction.live/stories/Fiction-liveBench-Mar-14-2025/oQdzQvKHw8JyXbN87/home", "sources": [ "https://epoch.ai/benchmarks/fictionlivebench" ], "unverified": [ "maintainer" ] }, { "id": "figure-unseen-homes-eval", "name": "Figure unseen-homes whole-task evaluation", "version": null, "category": "world models and robotics", "measures": "Vendor whole-task completion rate in unseen homes", "maintainer": "figure-ai", "status": "static", "url": null, "sources": [ "https://aiwiki.ai/wiki/helix_vla" ] }, { "id": "financebench", "name": "FinanceBench", "version": null, "category": "domain", "measures": "Financial document QA; 150 open cases of 10,231; launch paper GPT-4-Turbo with retrieval failed or refused about 81%", "maintainer": "patronus-ai", "status": null, "url": null, "sources": [ "https://capitalandcompute.net/ai-benchmarks/financebench/" ] }, { "id": "fiqa-sentiment", "name": "FiQA sentiment", "version": null, "category": "domain", "measures": "Financial sentiment analysis", "maintainer": null, "status": "active", "url": null, "sources": [ "https://arxiv.org/html/2305.05862v2" ] }, { "id": "fleurs", "name": "FLEURS", "version": null, "category": "speech", "measures": "Multilingual ASR error rate (FLEURS-102; Indonesian subset)", "maintainer": null, "status": "static", "url": null, "sources": [ "https://arxiv.org/pdf/2511.09690", "https://arxiv.org/html/2601.21337v2" ] }, { "id": "frontiercode-v1-1", "name": "FrontierCode", "version": "v1.1", "category": "coding", "measures": "Coding benchmark (Cognition's own bench per S8)", "maintainer": "cognition", "status": "active", "url": null, "sources": [ "https://llm-stats.com/blog/research/claude-opus-5-5-launch", "https://www.orcarouter.ai/fr/blog/swe-2" ] }, { "id": "frontiermath", "name": "FrontierMath", "version": "Tiers 1-4", "category": "math", "measures": "Several hundred unpublished, research-level maths problems; v2 reports Tiers 1-3 and Tier 4 separately; private Tier 4", "maintainer": "epoch-ai", "status": "active", "url": "https://epoch.ai/frontiermath", "sources": [ "https://arxiv.org/abs/2411.04872", "https://www.benchlm.ai/benchmarks/frontiermath", "https://epoch.ai/benchmarks" ], "unverified": [] }, { "id": "frontiermath-erdos", "name": "FrontierMath: Erdos", "version": null, "category": "math", "measures": "Open Erdos problems", "maintainer": "epoch-ai", "status": "active", "url": "https://epoch.ai/benchmarks", "sources": [ "https://epoch.ai/benchmarks" ], "unverified": [] }, { "id": "frontiermath-tier-4-v2", "name": "FrontierMath Tier 4", "version": "v2", "category": "math", "measures": "Hard research mathematics", "maintainer": null, "status": "active", "url": null, "sources": [ "https://pasqualepillitteri.it/en/news/14246/openai-launches-gpt-6-astra-pricing-benchmarks" ] }, { "id": "frontierswe", "name": "FrontierSWE", "version": null, "category": "coding", "measures": "Agentic software engineering", "maintainer": null, "status": "active", "url": null, "sources": [ "https://huggingface.co/moonshotai/Kimi-K3", "https://www.orcarouter.ai/es/blog/gemini-4-argon" ] }, { "id": "gaia", "name": "GAIA", "version": null, "category": "agents and tool use", "measures": "Real-world questions for general AI assistants that need tools and browsing", "maintainer": "Meta, Hugging Face and AutoGPT authors, USA/France", "status": "active", "url": "https://huggingface.co/spaces/gaia-benchmark/leaderboard", "sources": [ "https://arxiv.org/abs/2311.12983", "https://huggingface.co/spaces/gaia-benchmark/leaderboard", "https://benchlm.ai/benchmarks/gaia" ], "unverified": [ "maintainer", "status" ] }, { "id": "gdp-pdf", "name": "GDP.pdf", "version": null, "category": "knowledge and reasoning", "measures": "Component eval of the AA Intelligence Index v4.3.2", "maintainer": null, "status": "active", "url": null, "sources": [ "https://artificialanalysis.ai/" ], "unverified": [] }, { "id": "gdpval", "name": "GDPval", "version": null, "category": "agents and tool use", "measures": "Privately authored OpenAI benchmark", "maintainer": "openai", "status": "active", "url": null, "sources": [ "https://openai.com/index/why-we-no-longer-evaluate-swe-bench-verified/" ], "unverified": [] }, { "id": "gdpval-aa-v2-1", "name": "GDPval-AA", "version": "v2.1", "category": "agents and tool use", "measures": "Economically valuable task performance, Elo", "maintainer": "artificial-analysis", "status": "active", "url": null, "sources": [ "https://vpsranking.com/news/ai/ai-2026-09-22-anthropic-claude-opus-5-5/", "https://artificialanalysis.ai/" ] }, { "id": "gemini-robotics-2-vendor-evals", "name": "Gemini Robotics 2 vendor task evaluations", "version": null, "category": "world models and robotics", "measures": "Vendor task success rates (Apollo 2 whole-body pick, Franka Duo pick-and-place and insertion, ER 2 moment finding)", "maintainer": "google-deepmind", "status": "static", "url": null, "sources": [ "https://aiwiki.ai/wiki/gemini_robotics_2" ] }, { "id": "geneval", "name": "GenEval", "version": null, "category": "generative media", "measures": "Object-focused text-to-image compositional alignment: single object, two objects, counting, colours, position, colour attribution; scored with object detection", "maintainer": "GenEval authors (Ghosh, Hajishirzi, Schmidt)", "status": null, "url": "https://github.com/djghosh13/geneval", "sources": [ "https://github.com/djghosh13/geneval", "https://arxiv.org/abs/2310.11513" ] }, { "id": "genie-sim-3-0", "name": "Genie Sim 3.0 sim-to-real", "version": "3.0", "category": "world models and robotics", "measures": "Sim-to-real transfer success in AgiBot's Genie Sim 3.0", "maintainer": "agibot", "status": "active", "url": null, "sources": [ "https://www.therobotreport.com/agibot-releases-go-2-foundation-model-embodied-ai/" ] }, { "id": "geo-bench", "name": "GEO-Bench", "version": null, "category": "domain", "measures": "Geospatial foundation model evaluation", "maintainer": null, "status": "active", "url": null, "sources": [ "https://huggingface.co/ibm-nasa-geospatial/Prithvi-EO-2.0-600M", "https://arxiv.org/abs/2412.02732v2" ] }, { "id": "geo-bench-2", "name": "GEO-Bench-2", "version": null, "category": "domain", "measures": "Geospatial FM evaluation over 19 datasets", "maintainer": "AI Alliance", "status": "active", "url": null, "sources": [ "https://research.ibm.com/blog/earth-observation-leadership" ] }, { "id": "gift-eval", "name": "GIFT-Eval", "version": null, "category": "time series", "measures": "Probabilistic and point forecasting across 97 dataset configurations; MASE and CRPS normalised by Seasonal Naive", "maintainer": "salesforce", "status": "active", "url": "https://huggingface.co/spaces/Salesforce/GIFT-Eval", "sources": [ "https://huggingface.co/spaces/Salesforce/GIFT-Eval", "https://arxiv.org/pdf/2605.20119", "https://arxiv.org/abs/2410.10393", "https://www.amazon.science/blog/introducing-chronos-2-from-univariate-to-universal-forecasting", "https://datanorth.ai/nl/nieuws/google-lanceert-timesfm-3", "https://www.salesforce.com/blog/moirai-2-0/", "https://arxiv.org/html/2511.11698v1", "https://www.ibm.com/ae-ar/granite/docs/models/time-series/" ] }, { "id": "global-mmlu", "name": "Global MMLU", "version": null, "category": "knowledge and reasoning", "measures": "Global multilingual MMLU (SmolLM3 and Gemma 3n cards)", "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/HuggingFaceTB/SmolLM3-3B", "https://ai.google.dev/gemma/docs/gemma-3n/model_card" ], "unverified": [] }, { "id": "global-mmlu-lite", "name": "Global MMLU Lite", "version": null, "category": "Southeast Asian and Indonesian", "measures": "Multilingual knowledge QA", "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/aisingapore/Qwen-SEA-LION-v4.5-27B-IT" ] }, { "id": "gpqa", "name": "GPQA", "version": null, "category": "knowledge and reasoning", "measures": "Graduate-level Google-proof science questions (full set; subset not stated where cited)", "maintainer": "Rein et al. (NYU/Cohere/Anthropic authors), USA", "status": null, "url": "https://arxiv.org/abs/2311.12022", "sources": [ "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16", "https://huggingface.co/Qwen/Qwen3.5-397B-A17B", "https://arxiv.org/abs/2311.12022", "https://www.benchlm.ai/benchmarks/gpqa" ], "unverified": [ "maintainer" ] }, { "id": "gpqa-diamond", "name": "GPQA Diamond", "version": null, "category": "knowledge and reasoning", "measures": "Graduate-level science questions", "maintainer": "Rein et al. (NYU/Cohere/Anthropic authors), USA", "status": "crowded", "url": "https://arxiv.org/abs/2311.12022", "sources": [ "https://docs.apiyi.com/en/news/gpt-6-astra-launch", "https://pasqualepillitteri.it/en/news/14246/openai-launches-gpt-6-astra-pricing-benchmarks", "https://huggingface.co/moonshotai/Kimi-K3", "https://arxiv.org/abs/2311.12022", "https://www.benchlm.ai/benchmarks/gpqa", "https://epoch.ai/data-insights/llm-inference-price-trends" ], "unverified": [ "maintainer" ] }, { "id": "granite-guardian-ood-safety", "name": "Granite Guardian out-of-distribution safety set", "version": null, "category": "safety", "measures": "Out-of-distribution safety F1 as reported on the IBM card (dataset not named)", "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/ibm-granite/granite-guardian-4.1-8b" ] }, { "id": "graphwalks", "name": "Graphwalks", "version": null, "category": "long context", "measures": "Long-context reasoning", "maintainer": "openai", "status": "active", "url": null, "sources": [ "https://benchlm.ai/best/long-context" ], "unverified": [] }, { "id": "gsm8k", "name": "GSM8K", "version": null, "category": "math", "measures": "Grade-school maths word problems", "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/microsoft/Phi-4-mini-instruct" ], "unverified": [] }, { "id": "gsma-open-telco-llm-benchmarks", "name": "GSMA Open-Telco LLM Benchmarks", "version": null, "category": "domain", "measures": "Telecom LLM benchmarking initiative, open source; benchmarks not named on page", "maintainer": "gsma", "status": null, "url": "https://www.gsma.com/get-involved/gsma-foundry/gsma-open-telco-llm-benchmarks/", "sources": [ "https://www.gsma.com/get-involved/gsma-foundry/gsma-open-telco-llm-benchmarks/" ] }, { "id": "harmbench", "name": "HarmBench", "version": null, "category": "safety", "measures": "Red-team and refusal test", "maintainer": "Academic", "status": null, "url": null, "sources": [ "https://huggingface.co/mistralai/Shieldstral-1.0-3B" ], "unverified": [ "status" ] }, { "id": "harvey-legal-agent-benchmark", "name": "Harvey Legal Agent Benchmark", "version": null, "category": "domain", "measures": "Legal agent tasks", "maintainer": "Harvey", "status": "active", "url": null, "sources": [ "https://www.orcarouter.ai/es/blog/gemini-4-argon" ] }, { "id": "healthbench", "name": "HealthBench", "version": null, "category": "domain", "measures": "Health conversation quality", "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/baichuan-inc/Baichuan-M2-32B" ] }, { "id": "healthbench-consensus", "name": "HealthBench Consensus", "version": null, "category": "domain", "measures": "Consensus subset of HealthBench", "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/baichuan-inc/Baichuan-M2-32B" ] }, { "id": "healthbench-hard", "name": "HealthBench Hard", "version": null, "category": "domain", "measures": "Hard subset of HealthBench", "maintainer": null, "status": null, "url": null, "sources": [ "https://openai.com/index/introducing-gpt-5/", "https://huggingface.co/baichuan-inc/Baichuan-M2-32B" ] }, { "id": "hellaswag", "name": "HellaSwag", "version": null, "category": "knowledge and reasoning", "measures": "Commonsense completion (Gemma 3n PT card)", "maintainer": null, "status": null, "url": null, "sources": [ "https://ai.google.dev/gemma/docs/gemma-3n/model_card" ], "unverified": [] }, { "id": "hle", "name": "Humanity's Last Exam (HLE)", "version": null, "category": "knowledge and reasoning", "measures": "Expert-written closed questions at the frontier of knowledge, multimodal", "maintainer": "center-for-ai-safety and scale-ai", "status": "crowded", "url": "https://arxiv.org/abs/2501.14249", "sources": [ "https://huggingface.co/moonshotai/Kimi-K3", "https://arxiv.org/abs/2501.14249", "https://www.benchlm.ai/benchmarks/hle" ], "unverified": [ "maintainer" ] }, { "id": "humaneval", "name": "HumanEval", "version": null, "category": "coding", "measures": "Function-level code generation; saturated legacy metric", "maintainer": "openai", "status": "saturated", "url": null, "sources": [ "https://huggingface.co/ByteDance-Seed/Seed-Coder-8B-Instruct", "https://huggingface.co/bigcode/starcoder2-15b" ] }, { "id": "humaneval-plus", "name": "HumanEval+", "version": null, "category": "coding", "measures": "Extended HumanEval tests", "maintainer": "EvalPlus", "status": "saturated", "url": null, "sources": [ "https://huggingface.co/bigcode/starcoder2-15b", "https://huggingface.co/JetBrains/Mellum2.1-12B-A2.5B-Thinking" ] }, { "id": "humanitys-last-exam", "name": "Humanity's Last Exam", "version": null, "category": "knowledge and reasoning", "measures": "Expert-level questions", "maintainer": null, "status": "active", "url": "https://lastexam.ai/", "sources": [ "https://slash-digital.io/en/insights/ai-models-2026/", "https://lastexam.ai/" ] }, { "id": "idsv", "name": "IDSV (Indonesian Speech Variability)", "version": null, "category": "Southeast Asian and Indonesian", "measures": "Indonesian ASR WER/CER by speaking style (read or spontaneous, formal or informal, clean or noisy)", "maintainer": "institut-teknologi-bandung", "status": "static", "url": "https://arxiv.org/html/2410.08828v2", "sources": [ "https://arxiv.org/html/2410.08828v2" ] }, { "id": "ifbench", "name": "IFBench", "version": null, "category": "knowledge and reasoning", "measures": null, "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/ibm-granite/granite-4.2-8b" ] }, { "id": "ifeval", "name": "IFEval", "version": null, "category": "knowledge and reasoning", "measures": "Instruction following (SmolLM3 card)", "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/allenai/Olmo-3-32B-Think", "https://huggingface.co/HuggingFaceTB/SmolLM3-3B" ] }, { "id": "imagenet", "name": "ImageNet top-1", "version": null, "category": "domain", "measures": "Image classification accuracy (MCUNet repo)", "maintainer": null, "status": null, "url": null, "sources": [ "https://github.com/mit-han-lab/mcunet" ], "unverified": [] }, { "id": "indommlu", "name": "IndoMMLU", "version": null, "category": "Southeast Asian and Indonesian", "measures": "Indonesian school-exam questions, including local languages and culture", "maintainer": "MBZUAI authors (Koto et al.), UAE/Indonesia", "status": "static", "url": "https://arxiv.org/abs/2310.04928", "sources": [ "https://huggingface.co/Yellow-AI-NLP/komodo-7b-base", "https://arxiv.org/abs/2310.04928", "https://blogs.nvidia.com/?p=75648", "https://huggingface.co/GoToCompany/Llama-Sahabat-AI-v2-70B-IT" ] }, { "id": "indonlu", "name": "IndoNLU", "version": null, "category": "Southeast Asian and Indonesian", "measures": "12 Indonesian NLU tasks", "maintainer": "indonlp", "status": "static", "url": "https://github.com/IndoNLP/indonlu", "sources": [ "https://github.com/IndoNLP/indonlu", "https://huggingface.co/indobenchmark/indobert-base-p1", "https://arxiv.org/pdf/2009.05387" ], "unverified": [] }, { "id": "intphys-2", "name": "IntPhys 2", "version": "2", "category": "world models and robotics", "measures": "Physical plausibility", "maintainer": "meta", "status": "active", "url": "https://ai.meta.com/blog/v-jepa-2-world-model-benchmarks/", "sources": [ "https://ai.meta.com/blog/v-jepa-2-world-model-benchmarks/" ] }, { "id": "ioi", "name": "IOI", "version": null, "category": "coding", "measures": "International Olympiad in Informatics problems", "maintainer": null, "status": "active", "url": null, "sources": [ "https://vals.ai/models/anthropic_claude-fable-5-1" ] }, { "id": "kgc-safety", "name": "KGC-Safety", "version": null, "category": "safety", "measures": null, "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/LGAI-EXAONE/K-EXAONE-2.0-750B-A37B" ] }, { "id": "kimi-code-bench-v2", "name": "Kimi Code Bench", "version": "v2", "category": "coding", "measures": "Moonshot vendor coding benchmark", "maintainer": "moonshot-ai", "status": "active", "url": null, "sources": [ "https://huggingface.co/moonshotai/Kimi-K2.7-Code" ] }, { "id": "kmmlu-pro", "name": "KMMLU-Pro", "version": null, "category": "knowledge and reasoning", "measures": null, "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/upstage/Solar-Open2-250B", "https://huggingface.co/upstage" ] }, { "id": "labbench-2", "name": "LABBench 2", "version": "2", "category": "domain", "measures": "Laboratory and biology research tasks", "maintainer": null, "status": "active", "url": null, "sources": [ "https://www.orcarouter.ai/es/blog/gemini-4-argon" ] }, { "id": "libero", "name": "LIBERO", "version": null, "category": "world models and robotics", "measures": "Simulated robot manipulation task suites; average success rate", "maintainer": null, "status": "saturated", "url": null, "sources": [ "https://arxiv.org/abs/2502.19645", "https://www.therobotreport.com/agibot-releases-go-2-foundation-model-embodied-ai/" ], "unverified": [ "measures", "status" ] }, { "id": "libero-plus", "name": "LIBERO-Plus", "version": null, "category": "world models and robotics", "measures": "Harder LIBERO variant, increasingly cited as LIBERO saturates", "maintainer": null, "status": "active", "url": null, "sources": [ "https://www.therobotreport.com/agibot-releases-go-2-foundation-model-embodied-ai/" ] }, { "id": "libero-pro", "name": "LIBERO-PRO", "version": null, "category": "world models and robotics", "measures": "Harder LIBERO variant, increasingly cited as LIBERO saturates", "maintainer": null, "status": "active", "url": null, "sources": [ "https://www.therobotreport.com/agibot-releases-go-2-foundation-model-embodied-ai/" ] }, { "id": "librispeech", "name": "LibriSpeech", "version": null, "category": "speech", "measures": "English ASR WER (clean, other)", "maintainer": null, "status": "static", "url": null, "sources": [ "https://select.keenable.ai/r/speech-recognition-accuracy-leaders-2022-2026/JbURQjlVooLQ6J10NkbRVw" ] }, { "id": "livecodebench", "name": "LiveCodeBench", "version": null, "category": "coding", "measures": "Contamination-limited competitive programming problems", "maintainer": "UC Berkeley/MIT/Cornell authors, USA", "status": "active", "url": "https://arxiv.org/abs/2403.07974", "sources": [ "https://vals.ai/models/anthropic_claude-fable-5-1", "https://blog.buildfastwithai.com/deepseek-v4-pro-review-2026", "https://huggingface.co/Qwen/Qwen3.8-27B", "https://huggingface.co/JetBrains/Mellum2.1-12B-A2.5B-Thinking", "https://huggingface.co/ByteDance-Seed/Seed-Coder-8B-Instruct", "https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro", "https://arxiv.org/abs/2403.07974", "https://www.benchlm.ai/benchmarks/livecodebench", "https://benchlm.ai/coding" ], "unverified": [ "maintainer" ] }, { "id": "livecodebench-v5", "name": "LiveCodeBench v5", "version": "v5", "category": "coding", "measures": "LiveCodeBench version 5 (Gemma 3n card)", "maintainer": "UC Berkeley/MIT/Cornell authors, USA", "status": null, "url": "https://arxiv.org/abs/2403.07974", "sources": [ "https://ai.google.dev/gemma/docs/gemma-3n/model_card" ], "unverified": [] }, { "id": "livecodebench-v6", "name": "LiveCodeBench", "version": "v6", "category": "coding", "measures": null, "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/Qwen/Qwen3.8-Flash-Next" ] }, { "id": "lm-aggrefact", "name": "LM-AggreFact", "version": null, "category": "safety", "measures": null, "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/ibm-granite/granite-guardian-4.1-8b" ] }, { "id": "lmarena-agent", "name": "Arena Agent", "version": null, "category": "human preference arena", "measures": "Arena category leaderboard (human preference)", "maintainer": "lmarena", "status": "active", "url": "https://arena.ai/leaderboard", "sources": [ "https://arena.ai/leaderboard" ], "unverified": [] }, { "id": "lmarena-document", "name": "Arena Document", "version": null, "category": "human preference arena", "measures": "Arena category leaderboard (human preference)", "maintainer": "lmarena", "status": "active", "url": "https://arena.ai/leaderboard", "sources": [ "https://arena.ai/leaderboard" ], "unverified": [] }, { "id": "lmarena-image-edit", "name": "Arena Image Edit", "version": null, "category": "generative media", "measures": "Arena category leaderboard (human preference)", "maintainer": "lmarena", "status": "active", "url": "https://arena.ai/leaderboard", "sources": [ "https://arena.ai/leaderboard" ], "unverified": [] }, { "id": "lmarena-image-to-video", "name": "Arena Image-to-Video", "version": null, "category": "generative media", "measures": "Arena category leaderboard (human preference)", "maintainer": "lmarena", "status": "active", "url": "https://arena.ai/leaderboard", "sources": [ "https://arena.ai/leaderboard" ], "unverified": [] }, { "id": "lmarena-search", "name": "Arena Search", "version": null, "category": "human preference arena", "measures": "Arena category leaderboard (human preference)", "maintainer": "lmarena", "status": "active", "url": "https://arena.ai/leaderboard", "sources": [ "https://arena.ai/leaderboard" ], "unverified": [] }, { "id": "lmarena-text", "name": "Arena Text leaderboard", "version": null, "category": "human preference arena", "measures": "Crowd-sourced pairwise preference on text chat (arena.ai, formerly LMArena)", "maintainer": "lmarena", "status": "active", "url": "https://arena.ai/leaderboard/chat/text", "sources": [ "https://arena.ai/leaderboard/chat/text", "https://arena.ai/leaderboard/text", "https://arxiv.org/abs/2403.04132", "https://arxiv.org/abs/2504.20879", "https://arena.ai/leaderboard" ] }, { "id": "lmarena-text-to-image", "name": "LMArena Text-to-Image Arena", "version": null, "category": "generative media", "measures": "Arena score from pairwise human preference votes on text-to-image outputs", "maintainer": "lmarena", "status": "active", "url": "https://arena.ai/leaderboard/text-to-image", "sources": [ "https://arena.ai/leaderboard/text-to-image", "https://arena.ai/leaderboard" ] }, { "id": "lmarena-text-to-video", "name": "LMArena Text-to-Video Arena", "version": null, "category": "generative media", "measures": "Arena score from pairwise human preference votes on text-to-video outputs", "maintainer": "lmarena", "status": "active", "url": "https://arena.ai/leaderboard/text-to-video", "sources": [ "https://arena.ai/leaderboard/text-to-video", "https://arena.ai/leaderboard" ] }, { "id": "lmarena-video-edit", "name": "Arena Video Edit", "version": null, "category": "generative media", "measures": "Arena category leaderboard (human preference)", "maintainer": "lmarena", "status": "active", "url": "https://arena.ai/leaderboard", "sources": [ "https://arena.ai/leaderboard" ], "unverified": [] }, { "id": "lmarena-vision", "name": "Arena Vision leaderboard", "version": null, "category": "human preference arena", "measures": "Crowd-sourced pairwise preference on vision chat", "maintainer": "lmarena", "status": "active", "url": "https://arena.ai/leaderboard/chat/vision", "sources": [ "https://arena.ai/leaderboard/chat/vision", "https://arena.ai/leaderboard" ] }, { "id": "lmarena-webdev", "name": "Arena WebDev", "version": null, "category": "human preference arena", "measures": "Arena category leaderboard (human preference)", "maintainer": "lmarena", "status": "active", "url": "https://arena.ai/leaderboard", "sources": [ "https://arena.ai/leaderboard" ], "unverified": [] }, { "id": "longbench-v2", "name": "LongBench v2", "version": "v2", "category": "long context", "measures": "Retrieval and reasoning over long inputs", "maintainer": "Tsinghua", "status": "active", "url": null, "sources": [ "https://benchlm.ai/best/long-context" ], "unverified": [] }, { "id": "lvis", "name": "LVIS", "version": null, "category": "domain", "measures": "Instance segmentation and box detection (AP, cgF1)", "maintainer": null, "status": "static", "url": null, "sources": [ "https://github.com/facebookresearch/sam3" ] }, { "id": "matbench-discovery", "name": "Matbench Discovery", "version": null, "category": "domain", "measures": "MLIP materials discovery ranked by CPS; also F1 and kSRME", "maintainer": null, "status": "active", "url": "https://matbench-discovery.materialsproject.org/", "sources": [ "https://matbench-discovery.materialsproject.org/" ] }, { "id": "math", "name": "MATH", "version": null, "category": "math", "measures": "Competition maths problems (Phi-4-mini card)", "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/allenai/Olmo-3-32B-Think", "https://huggingface.co/microsoft/Phi-4-mini-instruct" ] }, { "id": "math-500", "name": "MATH500", "version": null, "category": "math", "measures": "500-problem maths subset", "maintainer": null, "status": "active", "url": null, "sources": [ "https://huggingface.co/LiquidAI/LFM2.5-8B-A1B", "https://huggingface.co/microsoft/Phi-4-mini-flash-reasoning", "https://arxiv.org/abs/2502.03387" ] }, { "id": "mbpp", "name": "MBPP", "version": null, "category": "coding", "measures": "Basic Python problems; saturated legacy metric", "maintainer": null, "status": "saturated", "url": null, "sources": [ "https://huggingface.co/ByteDance-Seed/Seed-Coder-8B-Instruct" ] }, { "id": "mbpp-plus", "name": "MBPP+", "version": null, "category": "coding", "measures": "Extended MBPP tests", "maintainer": "EvalPlus", "status": "saturated", "url": null, "sources": [ "https://huggingface.co/JetBrains/Mellum2.1-12B-A2.5B-Thinking" ] }, { "id": "mcp-atlas", "name": "MCP Atlas", "version": null, "category": "agents and tool use", "measures": "Agent benchmark tracked by BenchLM", "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/meta-models/Muse-Glimmer-30B", "https://www.benchlm.ai/benchmarks" ] }, { "id": "medagentbench", "name": "MedAgentBench", "version": null, "category": "domain", "measures": "Medical agent tasks; Claude scores shown only in charts", "maintainer": null, "status": null, "url": null, "sources": [ "https://www.anthropic.com/news/healthcare-life-sciences" ] }, { "id": "medcalc", "name": "MedCalc", "version": null, "category": "domain", "measures": "Medical calculation; Claude scores shown only in charts", "maintainer": null, "status": null, "url": null, "sources": [ "https://www.anthropic.com/news/healthcare-life-sciences" ] }, { "id": "medmcqa", "name": "MedMCQA", "version": null, "category": "domain", "measures": "Medical multiple-choice QA", "maintainer": null, "status": null, "url": null, "sources": [ "https://arxiv.org/html/2507.05201v2", "https://huggingface.co/m42-health/Llama3-Med42-70B" ] }, { "id": "medqa", "name": "MedQA", "version": null, "category": "domain", "measures": "Medical licensing-style multiple-choice QA", "maintainer": null, "status": null, "url": null, "sources": [ "https://arxiv.org/html/2507.05201v2", "https://huggingface.co/google/medgemma-27b-text-it", "https://huggingface.co/m42-health/Llama3-Med42-70B" ] }, { "id": "metr-time-horizon", "name": "METR task time horizon", "version": null, "category": "agents and tool use", "measures": "Length of tasks models complete at 50% reliability; doubling about every 7 months", "maintainer": "metr", "status": "active", "url": "https://metr.org/blog/2025-03-19-measuring-ai-ability-to-complete-long-tasks/", "sources": [ "https://metr.org/blog/2025-03-19-measuring-ai-ability-to-complete-long-tasks/" ] }, { "id": "mgsm", "name": "MGSM", "version": null, "category": "math", "measures": "Multilingual grade-school maths (Gemma 3n card)", "maintainer": null, "status": null, "url": null, "sources": [ "https://ai.google.dev/gemma/docs/gemma-3n/model_card" ], "unverified": [] }, { "id": "mimic-cxr", "name": "MIMIC-CXR", "version": null, "category": "domain", "measures": "Chest X-ray classification (macro F1) and findings generation", "maintainer": null, "status": null, "url": null, "sources": [ "https://arxiv.org/html/2507.05201v2", "https://www.microsoft.com/en-us/research/publication/maira-2-grounded-radiology-report-generation/" ] }, { "id": "mlperf-inference-edge-agentic", "name": "MLPerf Edge Agentic Inference", "version": "v6.1", "category": "efficiency", "measures": "Multi-turn agentic tasks (e.g. coding) on edge devices with an edge model and quantization; model not named", "maintainer": "mlcommons", "status": "active", "url": "https://mlcommons.org/benchmarks/inference-edge/", "sources": [ "https://www.financialcontent.com/article/gnwcq-2026-9-16-mlcommons-sets-participation-record-with-new-mlperf-inference-v61-benchmark-results", "https://mlcommons.org/benchmarks/inference-edge/" ], "unverified": [] }, { "id": "mlperf-inference-v5-1", "name": "MLPerf Inference", "version": "v5.1", "category": "efficiency", "measures": "Throughput and latency at a quality target; Sep 2025 round added Llama 3.2 1B and 3B for edge", "maintainer": "mlcommons", "status": "superseded", "url": null, "sources": [ "https://www.winzheng.com/en/article/mlperf-inference-v5-1-results" ], "unverified": [] }, { "id": "mlperf-inference-v6-0", "name": "MLPerf Inference", "version": "v6.0", "category": "efficiency", "measures": "1 Apr 2026 round, 24 submitters; upgraded edge object detection on Ultralytics YOLOv11 Large plus a new edge object-detection test", "maintainer": "mlcommons", "status": "superseded", "url": null, "sources": [ "https://www.globenewswire.com/news-release/2026/04/01/3266801/0/en/mlcommons-releases-new-mlperf-inference-v6-0-benchmark-results.html" ], "unverified": [] }, { "id": "mlperf-inference-v6-1", "name": "MLPerf Inference (Datacenter/Edge)", "version": "v6.1", "category": "efficiency", "measures": "16 Sep 2026 round, record 30 submitters; new Edge Agentic Inference benchmark", "maintainer": "mlcommons", "status": "active", "url": "https://mlcommons.org/benchmarks/inference-datacenter/", "sources": [ "https://www.financialcontent.com/article/gnwcq-2026-9-16-mlcommons-sets-participation-record-with-new-mlperf-inference-v61-benchmark-results", "https://mlcommons.org/benchmarks/inference-edge/", "https://mlcommons.org/benchmarks/inference-datacenter/" ], "unverified": [] }, { "id": "mlperf-tiny-v1-3", "name": "MLPerf Tiny", "version": "v1.3", "category": "efficiency", "measures": "Inference on microcontrollers; 17 Sep 2025 round added streaming wakeword (1D DS-CNN); 70 results, 27 power results; submitters Kai Jiang, Qualcomm, STMicroelectronics, Syntiant", "maintainer": "mlcommons", "status": "superseded", "url": "https://mlcommons.org/benchmarks/inference-tiny/", "sources": [ "https://www.globenewswire.com/news-release/2025/09/17/3151817/0/en/MLCommons-New-MLPerf-Tiny-v1-3-Benchmark-Results-Released.html", "https://mlcommons.org/benchmarks/inference-tiny/" ], "unverified": [] }, { "id": "mlperf-tiny-v1-4", "name": "MLPerf Tiny", "version": "v1.4", "category": "efficiency", "measures": "Inference on microcontrollers; same five tasks as v1.3; release date not stated (page modified 2026-07-07)", "maintainer": "mlcommons", "status": "active", "url": "https://mlcommons.org/benchmarks/inference-tiny/", "sources": [ "https://mlcommons.org/benchmarks/inference-tiny/" ], "unverified": [ "version" ] }, { "id": "mlperf-training-v6-0", "name": "MLPerf Training", "version": "v6.0", "category": "efficiency", "measures": "Time to train to a quality target. Workloads: DeepSeek v3, GPT-OSS 20B, Llama 3.1 8B/405B pretraining; Llama 2 70B fine-tuning; DLRM-dcnv2; FLUX.1", "maintainer": "mlcommons", "status": "active", "url": "https://mlcommons.org/benchmarks/training/", "sources": [ "https://mlcommons.org/benchmarks/training/" ], "unverified": [] }, { "id": "mmlongbench-doc", "name": "MMLongBench-Doc", "version": null, "category": "long context", "measures": "Long-document multimodal understanding", "maintainer": null, "status": "active", "url": null, "sources": [ "https://benchlm.ai/best/long-context" ], "unverified": [] }, { "id": "mmlu", "name": "MMLU", "version": null, "category": "knowledge and reasoning", "measures": "Multiple-choice knowledge questions (reported 5-shot on the Phi-4-mini card)", "maintainer": null, "status": "saturated", "url": null, "sources": [ "https://huggingface.co/microsoft/Phi-4-mini-instruct", "https://arxiv.org/abs/2501.14249", "https://arxiv.org/abs/2404.14219" ], "unverified": [] }, { "id": "mmlu-med", "name": "MMLU Med", "version": null, "category": "domain", "measures": "Medical subsets of MMLU", "maintainer": null, "status": null, "url": null, "sources": [ "https://arxiv.org/html/2507.05201v2" ] }, { "id": "mmlu-pro", "name": "MMLU-Pro", "version": null, "category": "knowledge and reasoning", "measures": "10-option multiple-choice questions across 14 domains; a harder version of MMLU", "maintainer": "TIGER-Lab (U. Waterloo et al.), Canada", "status": "crowded", "url": "https://arxiv.org/abs/2406.01574", "sources": [ "https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro", "https://arxiv.org/abs/2406.01574", "https://www.benchlm.ai/benchmarks/mmlu-pro" ], "unverified": [ "maintainer" ] }, { "id": "mmmu", "name": "MMMU", "version": null, "category": "multimodal", "measures": "College-level multi-discipline questions over images", "maintainer": "MMMU team (OSU/Waterloo et al.), USA/Canada", "status": "active", "url": "https://arxiv.org/abs/2311.16502", "sources": [ "https://huggingface.co/microsoft/Phi-4-reasoning-vision-15B", "https://arxiv.org/abs/2311.16502", "https://www.benchlm.ai/benchmarks/mmmu" ], "unverified": [ "maintainer" ] }, { "id": "mmmu-pro", "name": "MMMU-Pro", "version": null, "category": "multimodal", "measures": "Harder MMMU variant", "maintainer": "MMMU team", "status": null, "url": null, "sources": [ "https://huggingface.co/MiniMaxAI/MiniMax-M3", "https://www.benchlm.ai/benchmarks/mmmu", "https://www.artificialanalysis.ai/articles/openbmb-launches-minicpm-v-4-6-1-3b-instruct" ] }, { "id": "mmteb", "name": "MMTEB", "version": null, "category": "embeddings", "measures": null, "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/nvidia/Nemotron-3-Embed-8B-BF16", "https://huggingface.co/jinaai/jina-embeddings-v5-text-small" ] }, { "id": "mmteb-r", "name": "MMTEB-R (reranking)", "version": null, "category": "embeddings", "measures": null, "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/Qwen/Qwen3-Reranker-8B" ] }, { "id": "molmospaces", "name": "MolmoSpaces", "version": null, "category": "world models and robotics", "measures": "Robot policy evaluation cited by analysts for GR00T N1.7", "maintainer": null, "status": "active", "url": null, "sources": [ "https://www.originofbots.com/news/nvidia-gr00t-n17-hits-early-access-for-humanoids" ] }, { "id": "mrcr-v2", "name": "MRCR v2", "version": "v2", "category": "long context", "measures": "Long-context retrieval and reasoning", "maintainer": "openai", "status": "active", "url": null, "sources": [ "https://benchlm.ai/best/long-context" ], "unverified": [] }, { "id": "mt-bench", "name": "MT-Bench", "version": null, "category": "knowledge and reasoning", "measures": "Multi-turn chat quality scored 1 to 10 by an LLM judge", "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/nvidia/Nemotron-4-340B-Instruct", "https://qwenlm.github.io/blog/qwen1.5-110b/", "https://mistral.ai/news/la-plateforme/" ] }, { "id": "mteb", "name": "MTEB / MMTEB", "version": "v4", "category": "embeddings", "measures": "Embedding quality: 500+ tasks in 250+ languages (retrieval, classification, clustering and more)", "maintainer": "MTEB community (Enevoldsen, Muennighoff et al.), Denmark/intl", "status": "active", "url": "https://huggingface.co/spaces/mteb/leaderboard", "sources": [ "https://arxiv.org/abs/2502.13595", "https://huggingface.co/spaces/mteb/leaderboard", "https://github.com/embeddings-benchmark/mteb" ], "unverified": [] }, { "id": "mteb-code", "name": "MTEB Code", "version": null, "category": "embeddings", "measures": null, "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/google/embeddinggemma-2" ] }, { "id": "mteb-english-v2", "name": "MTEB English", "version": "v2", "category": "embeddings", "measures": null, "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/google/embeddinggemma-2", "https://huggingface.co/jinaai/jina-embeddings-v5-text-small" ] }, { "id": "mteb-multilingual", "name": "MTEB multilingual", "version": null, "category": "embeddings", "measures": null, "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/Qwen/Qwen3-Embedding-8B" ] }, { "id": "mteb-multilingual-v2", "name": "MTEB multilingual", "version": "v2", "category": "embeddings", "measures": null, "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/google/embeddinggemma-2" ] }, { "id": "mteb-r", "name": "MTEB-R (reranking)", "version": null, "category": "embeddings", "measures": null, "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/Qwen/Qwen3-Reranker-8B" ] }, { "id": "multijail", "name": "MultiJail", "version": null, "category": "safety", "measures": null, "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/nvidia/Nemotron-3.5-Content-Safety" ] }, { "id": "multilingual-mmlu", "name": "Multilingual MMLU", "version": null, "category": "knowledge and reasoning", "measures": "Multilingual version of MMLU (Phi-4-mini card)", "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/microsoft/Phi-4-mini-instruct" ], "unverified": [] }, { "id": "mvpbench", "name": "MVPBench", "version": null, "category": "world models and robotics", "measures": "Minimal-pair video QA", "maintainer": "meta", "status": "active", "url": "https://ai.meta.com/blog/v-jepa-2-world-model-benchmarks/", "sources": [ "https://ai.meta.com/blog/v-jepa-2-world-model-benchmarks/" ] }, { "id": "mvtec-ad", "name": "MVTec AD", "version": null, "category": "industrial", "measures": "Unsupervised anomaly detection and localization, 5,000+ images, 15 categories; CC BY-NC-SA 4.0", "maintainer": "mvtec", "status": "saturated", "url": "https://www.mvtec.com/company/research/datasets/mvtec-ad", "sources": [ "https://www.mvtec.com/company/research/datasets/mvtec-ad", "https://arxiv.org/pdf/2510.17611" ], "unverified": [ "status" ] }, { "id": "mvtec-ad-2", "name": "MVTec AD 2", "version": "2", "category": "industrial", "measures": "Successor anomaly-detection dataset", "maintainer": "mvtec", "status": null, "url": null, "sources": [ "https://www.mvtec.com/company/research/datasets/mvtec-ad" ], "unverified": [] }, { "id": "n-cmapss", "name": "N-CMAPSS", "version": "2021", "category": "industrial", "measures": "Newer C-MAPSS dataset (2021) with ETH Zurich and PARC", "maintainer": "nasa", "status": null, "url": "https://www.nasa.gov/intelligent-systems-division/discovery-and-systems-health/pcoe/pcoe-data-set-repository/", "sources": [ "https://www.nasa.gov/intelligent-systems-division/discovery-and-systems-health/pcoe/pcoe-data-set-repository/" ], "unverified": [] }, { "id": "nab", "name": "NAB (Numenta Anomaly Benchmark)", "version": "v1.0", "category": "time series", "measures": "Streaming univariate anomaly detection; legacy, little maintenance", "maintainer": "Numenta", "status": "static", "url": "https://github.com/numenta/NAB", "sources": [ "https://github.com/numenta/NAB" ] }, { "id": "neuco-bench", "name": "NeuCo-bench", "version": null, "category": "domain", "measures": "Neural compression for EO", "maintainer": null, "status": "active", "url": null, "sources": [ "https://research.ibm.com/blog/earth-observation-leadership" ] }, { "id": "nodeground", "name": "NodeGround", "version": null, "category": "domain", "measures": "Node-classification benchmark for graph foundation models", "maintainer": null, "status": "active", "url": "https://arxiv.org/abs/2609.39673", "sources": [ "https://arxiv.org/abs/2609.39673" ] }, { "id": "nusacrowd", "name": "NusaCrowd", "version": null, "category": "Southeast Asian and Indonesian", "measures": "137 datasets with 118 standardised loaders; first zero-shot benchmarks and first multilingual ASR benchmark for Indonesian languages (2022, v4 2023)", "maintainer": null, "status": "active", "url": "https://arxiv.org/abs/2212.09648", "sources": [ "https://arxiv.org/abs/2212.09648" ] }, { "id": "nusax", "name": "NusaX", "version": null, "category": "Southeast Asian and Indonesian", "measures": "Parallel sentiment and MT corpus: Indonesian, English and 10 local languages", "maintainer": "indonlp", "status": "static", "url": "https://github.com/IndoNLP/nusax", "sources": [ "https://github.com/IndoNLP/nusax", "https://arxiv.org/abs/2205.15960" ], "unverified": [] }, { "id": "olmocr-bench", "name": "olmOCR-Bench", "version": null, "category": "multimodal", "measures": "OCR benchmark from Ai2", "maintainer": "allen-ai", "status": "active", "url": "https://benchmarklist.com/benchmarks/olmocr_bench/", "sources": [ "https://benchmarklist.com/benchmarks/olmocr_bench/" ] }, { "id": "omnidocbench-v1-6", "name": "OmniDocBench", "version": "v1.6", "category": "multimodal", "measures": "Document parsing overall score, higher is better", "maintainer": null, "status": "active", "url": null, "sources": [ "https://arxiv.org/pdf/2606.03264" ] }, { "id": "open-asr-leaderboard", "name": "Hugging Face Open ASR Leaderboard", "version": null, "category": "speech", "measures": "Eight-set macro-average English WER, lower is better; mostly read and broadcast English", "maintainer": "hugging-face", "status": "crowded", "url": "https://benchmarklist.com/benchmarks/open_asr_leaderboard/", "sources": [ "https://benchmarklist.com/benchmarks/open_asr_leaderboard/", "https://www.gladia.io/blog/best-open-source-speech-to-text-models", "https://huggingface.co/spaces/hf-audio/open_asr_leaderboard" ] }, { "id": "open-x-embodiment-evaluations", "name": "Open X-Embodiment evaluations", "version": null, "category": "world models and robotics", "measures": "Cross-lab evaluations of RT-1-X and RT-2-X robot policies trained on the Open X-Embodiment dataset (1M+ trajectories, 22 embodiments)", "maintainer": "Open X-Embodiment Collaboration", "status": null, "url": "https://robotics-transformer-x.github.io/", "sources": [ "https://robotics-transformer-x.github.io/" ] }, { "id": "openai-mrcr-v2", "name": "OpenAI MRCR", "version": "v2", "category": "long context", "measures": "Multi-round coreference retrieval at 512K to 1M tokens", "maintainer": "openai", "status": "active", "url": null, "sources": [ "https://slash-digital.io/en/insights/ai-models-2026/" ] }, { "id": "osworld", "name": "OSWorld", "version": null, "category": "agents and tool use", "measures": "Computer use by an agent operating a desktop from screenshots", "maintainer": null, "status": null, "url": null, "sources": [ "https://www.anthropic.com/news/3-5-models-and-computer-use" ] }, { "id": "osworld-2-0", "name": "OSWorld", "version": "2.0", "category": "agents and tool use", "measures": "Computer-use tasks", "maintainer": "xlang-lab-hku", "status": "active", "url": null, "sources": [ "https://vpsranking.com/news/ai/ai-2026-09-22-anthropic-claude-opus-5-5/", "https://pasqualepillitteri.it/en/news/14246/openai-launches-gpt-6-astra-pricing-benchmarks", "https://www.benchlm.ai/benchmarks/osworld-verified" ] }, { "id": "osworld-verified", "name": "OSWorld-Verified", "version": null, "category": "agents and tool use", "measures": "Computer-use tasks", "maintainer": "xlang-lab-hku", "status": "crowded", "url": "https://arxiv.org/abs/2404.07972", "sources": [ "https://openai.com/index/introducing-gpt-5-3-codex", "https://huggingface.co/moonshotai/Kimi-K3", "https://arxiv.org/abs/2404.07972", "https://www.benchlm.ai/benchmarks/osworld-verified" ], "unverified": [ "maintainer" ] }, { "id": "pai-bench", "name": "PAI-Bench", "version": null, "category": "world models and robotics", "measures": "Physical-AI world generation quality (cited by NVIDIA for Cosmos 3)", "maintainer": null, "status": "active", "url": null, "sources": [ "https://investor.nvidia.com/news/press-release-details/2026/NVIDIA-Launches-Cosmos-3-the-Open-Frontier-Foundation-Model-for-Physical-AI/default.aspx" ] }, { "id": "pangaea", "name": "PANGAEA", "version": null, "category": "domain", "measures": "Geospatial FM evaluation", "maintainer": null, "status": "active", "url": null, "sources": [ "https://research.ibm.com/blog/earth-observation-leadership" ] }, { "id": "perceptiontest", "name": "PerceptionTest", "version": null, "category": "multimodal", "measures": "Video understanding QA", "maintainer": null, "status": "active", "url": null, "sources": [ "https://arxiv.org/abs/2506.09985" ] }, { "id": "phm-society-data-challenges", "name": "PHM Society Data Challenges", "version": null, "category": "industrial", "measures": "Various prognostics challenges; 2024/2025 pages failed to load", "maintainer": null, "status": null, "url": null, "sources": [ "https://arxiv.org/abs/2509.22267v1" ] }, { "id": "physics-iq", "name": "Physics-IQ", "version": null, "category": "world models and robotics", "measures": "Physics accuracy of world generation (cited by NVIDIA for Cosmos 3)", "maintainer": null, "status": "active", "url": null, "sources": [ "https://investor.nvidia.com/news/press-release-details/2026/NVIDIA-Launches-Cosmos-3-the-Open-Frontier-Foundation-Model-for-Physical-AI/default.aspx" ] }, { "id": "phyworld", "name": "PhyWorld", "version": null, "category": "world models and robotics", "measures": "Whether video generators learn physical laws in-, combinatorial- and out-of-distribution", "maintainer": "Kang et al. (ICML 2025 submission)", "status": "static", "url": "https://arxiv.org/abs/2411.02385", "sources": [ "https://arxiv.org/abs/2411.02385" ] }, { "id": "polyguard", "name": "PolyGuard", "version": null, "category": "safety", "measures": null, "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/mistralai/Shieldstral-1.0-3B", "https://huggingface.co/nvidia/Nemotron-3.5-Content-Safety" ] }, { "id": "posebusters", "name": "PoseBusters", "version": null, "category": "domain", "measures": "Protein-ligand pose prediction validity", "maintainer": null, "status": null, "url": null, "sources": [ "https://www.nature.com/articles/s41586-024-07487-w" ] }, { "id": "programbench", "name": "ProgramBench", "version": null, "category": "coding", "measures": "Program construction tasks", "maintainer": null, "status": "active", "url": null, "sources": [ "https://vals.ai/models/anthropic_claude-fable-5-1", "https://huggingface.co/moonshotai/Kimi-K3", "https://huggingface.co/moonshotai/Kimi-K2.7-Code" ] }, { "id": "pronostia-femto", "name": "PRONOSTIA / FEMTO (IEEE PHM 2012)", "version": null, "category": "industrial", "measures": "Bearing remaining useful life; no current SOTA retrieved", "maintainer": null, "status": null, "url": null, "sources": [ "https://arxiv.org/abs/2509.22267v1" ] }, { "id": "protocol-qa", "name": "Protocol QA", "version": null, "category": "domain", "measures": "Life-science lab protocol QA; human baseline 0.79", "maintainer": null, "status": null, "url": null, "sources": [ "https://www.anthropic.com/news/claude-for-life-sciences" ] }, { "id": "pubmedqa", "name": "PubMedQA", "version": null, "category": "domain", "measures": "Biomedical literature QA", "maintainer": null, "status": null, "url": null, "sources": [ "https://arxiv.org/html/2507.05201v2" ] }, { "id": "qwen-image-bench", "name": "Qwen-Image-Bench", "version": null, "category": "generative media", "measures": "Qwen vendor image benchmark; methodology unreleased, chart only", "maintainer": "alibaba", "status": "static", "url": null, "sources": [ "https://datanorth.ai/news/qwen-releases-qwen-image-2-1" ] }, { "id": "r-bench", "name": "R-Bench", "version": null, "category": "world models and robotics", "measures": "World generation quality (cited by NVIDIA for Cosmos 3)", "maintainer": null, "status": "active", "url": null, "sources": [ "https://investor.nvidia.com/news/press-release-details/2026/NVIDIA-Launches-Cosmos-3-the-Open-Frontier-Foundation-Model-for-Physical-AI/default.aspx" ] }, { "id": "radgraph-f1", "name": "RadGraph F1", "version": null, "category": "domain", "measures": "Radiology report generation quality", "maintainer": null, "status": null, "url": null, "sources": [ "https://arxiv.org/html/2507.05201v2" ] }, { "id": "real-iad", "name": "Real-IAD", "version": null, "category": "industrial", "measures": "Industrial image anomaly detection; harder, more headroom", "maintainer": null, "status": "active", "url": null, "sources": [ "https://arxiv.org/pdf/2510.17611" ] }, { "id": "real5-omnidocbench", "name": "Real5-OmniDocBench", "version": null, "category": "multimodal", "measures": "Document parsing on real-world captures", "maintainer": null, "status": "active", "url": null, "sources": [ "https://arxiv.org/pdf/2606.03264" ] }, { "id": "recif-bench", "name": "RecIF-Bench", "version": null, "category": "domain", "measures": "Recommendation: 100M interactions, 200k users, 8 tasks; short video, ads, product", "maintainer": "kuaishou", "status": "active", "url": null, "sources": [ "https://github.com/Kuaishou-OneRec/OpenOneRec" ] }, { "id": "roboarena", "name": "RoboArena", "version": null, "category": "world models and robotics", "measures": "Robot policy evaluation (cited by NVIDIA for Cosmos 3 and by analysts for GR00T N1.7)", "maintainer": null, "status": "active", "url": null, "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://www.originofbots.com/news/nvidia-gr00t-n17-hits-early-access-for-humanoids" ], "unverified": [ "measures", "status" ] }, { "id": "robolab", "name": "RoboLab", "version": null, "category": "world models and robotics", "measures": "Robot action prediction (cited by NVIDIA for Cosmos 3)", "maintainer": null, "status": "active", "url": null, "sources": [ "https://investor.nvidia.com/news/press-release-details/2026/NVIDIA-Launches-Cosmos-3-the-Open-Frontier-Foundation-Model-for-Physical-AI/default.aspx" ] }, { "id": "rteb", "name": "RTEB", "version": null, "category": "embeddings", "measures": null, "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/nvidia/Nemotron-3-Embed-8B-BF16" ] }, { "id": "ruler", "name": "RULER", "version": null, "category": "long context", "measures": "Synthetic long-context tests", "maintainer": "nvidia", "status": null, "url": null, "sources": [ "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16" ], "unverified": [ "status", "maintainer" ] }, { "id": "sa-co-gold", "name": "SA-Co/Gold", "version": null, "category": "domain", "measures": "Promptable concept segmentation, cgF1", "maintainer": "meta", "status": "active", "url": null, "sources": [ "https://github.com/facebookresearch/sam3" ] }, { "id": "sa-v", "name": "SA-V", "version": null, "category": "domain", "measures": "Video segmentation, cgF1", "maintainer": "meta", "status": "active", "url": null, "sources": [ "https://github.com/facebookresearch/sam3" ] }, { "id": "scicode", "name": "SciCode", "version": null, "category": "knowledge and reasoning", "measures": "Component eval of the AA Intelligence Index v4.3.2", "maintainer": null, "status": "active", "url": null, "sources": [ "https://artificialanalysis.ai/" ], "unverified": [] }, { "id": "sea-helm", "name": "SEA-HELM", "version": null, "category": "Southeast Asian and Indonesian", "measures": "SEA average and per-language (incl. Indonesian) scores for open-weight models", "maintainer": "ai-singapore", "status": "active", "url": "https://leaderboard.sea-lion.ai/detailed/SEA", "sources": [ "https://leaderboard.sea-lion.ai/detailed/SEA", "https://leaderboard.sea-lion.ai/detailed/ID", "https://leaderboard.sea-lion.ai/", "https://arxiv.org/abs/2502.14301", "https://blogs.nvidia.com/?p=75648", "https://huggingface.co/GoToCompany/Llama-Sahabat-AI-v2-70B-IT", "https://huggingface.co/aisingapore/Qwen-SEA-LION-v4.5-27B-IT" ] }, { "id": "sea-ifeval", "name": "SEA-IFEval", "version": null, "category": "Southeast Asian and Indonesian", "measures": "Instruction following in SEA languages", "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/GoToCompany/Llama-Sahabat-AI-v2-70B-IT", "https://huggingface.co/aisingapore/Qwen-SEA-LION-v4.5-27B-IT" ] }, { "id": "sea-mtbench", "name": "SEA-MTBench", "version": null, "category": "Southeast Asian and Indonesian", "measures": "Multi-turn chat in SEA languages", "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/aisingapore/Qwen-SEA-LION-v4.5-27B-IT" ] }, { "id": "sea-safeguardbench", "name": "SEA-SafeguardBench", "version": null, "category": "safety", "measures": "Guard model performance on SEA languages (AUPRC metric); arXiv 2512.05501", "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/aisingapore/Qwen-SEA-Guard-8B-2602" ] }, { "id": "seabench", "name": "SeaBench", "version": null, "category": "Southeast Asian and Indonesian", "measures": "Multi-turn open tasks from SEA daily life", "maintainer": "DAMO Academy (Alibaba) SeaLLMs team, Singapore/China", "status": "active", "url": "https://huggingface.co/spaces/SeaLLMs/SeaExam_leaderboard", "sources": [ "https://arxiv.org/abs/2502.06298" ], "unverified": [] }, { "id": "seaexam", "name": "SeaExam", "version": null, "category": "Southeast Asian and Indonesian", "measures": "Regional school exams", "maintainer": "DAMO Academy (Alibaba) SeaLLMs team, Singapore/China", "status": "active", "url": "https://huggingface.co/spaces/SeaLLMs/SeaExam_leaderboard", "sources": [ "https://arxiv.org/abs/2502.06298", "https://huggingface.co/spaces/SeaLLMs/SeaExam_leaderboard" ], "unverified": [] }, { "id": "simplerenv", "name": "SimplerEnv", "version": null, "category": "world models and robotics", "measures": "Simulated evaluation of real-robot manipulation policies; how well simulation predicts real-world performance (MMRV, Pearson correlation)", "maintainer": "UC San Diego, Stanford, UC Berkeley and Google DeepMind researchers", "status": null, "url": "https://simpler-env.github.io/", "sources": [ "https://simpler-env.github.io/" ] }, { "id": "skab", "name": "SKAB", "version": null, "category": "time series", "measures": "Skoltech pump testbed multivariate anomaly detection (accelerometers, current, pressure, temperature); outlier F1 and changepoint NAB score; leaderboard undated", "maintainer": "Skoltech", "status": null, "url": "https://github.com/waico/SKAB", "sources": [ "https://github.com/waico/SKAB" ] }, { "id": "ssv2", "name": "Something-Something v2", "version": null, "category": "multimodal", "measures": "Video action recognition, top-1 accuracy", "maintainer": null, "status": "active", "url": null, "sources": [ "https://arxiv.org/abs/2506.09985" ] }, { "id": "strongreject", "name": "StrongREJECT", "version": null, "category": "safety", "measures": "Jailbreak effectiveness: how much useful information a model gives in response to forbidden prompts", "maintainer": null, "status": null, "url": "https://arxiv.org/abs/2402.10260", "sources": [ "https://arxiv.org/abs/2402.10260" ] }, { "id": "swe-bench-bash-only", "name": "SWE-bench Bash Only", "version": null, "category": "coding", "measures": "SWE-bench subset or variant", "maintainer": "SWE-bench team, USA", "status": "active", "url": "https://www.swebench.com/", "sources": [ "https://www.swebench.com/" ], "unverified": [] }, { "id": "swe-bench-full", "name": "SWE-bench Full", "version": null, "category": "coding", "measures": "SWE-bench subset or variant", "maintainer": "SWE-bench team, USA", "status": "active", "url": "https://www.swebench.com/", "sources": [ "https://www.swebench.com/" ], "unverified": [] }, { "id": "swe-bench-lite", "name": "SWE-bench Lite", "version": null, "category": "coding", "measures": "SWE-bench subset or variant", "maintainer": "SWE-bench team, USA", "status": "active", "url": "https://www.swebench.com/", "sources": [ "https://www.swebench.com/" ], "unverified": [] }, { "id": "swe-bench-multilingual", "name": "SWE-bench Multilingual", "version": null, "category": "coding", "measures": "Multilingual SWE-bench", "maintainer": "SWE-bench team, USA", "status": "active", "url": "https://www.swebench.com/", "sources": [ "https://cursor.com/blog/composer-2", "https://huggingface.co/mistralai/Devstral-2-123B-Instruct-2512", "https://www.swebench.com/" ] }, { "id": "swe-bench-multimodal", "name": "SWE-bench Multimodal", "version": null, "category": "coding", "measures": "SWE-bench subset or variant", "maintainer": "SWE-bench team, USA", "status": "active", "url": "https://www.swebench.com/", "sources": [ "https://www.swebench.com/" ], "unverified": [] }, { "id": "swe-bench-pro", "name": "SWE-bench Pro", "version": null, "category": "coding", "measures": "Software engineering tasks; vendor figures are harness-specific and OpenAI claims about 30% of tasks are broken", "maintainer": "scale-ai", "status": "active", "url": "https://labs.scale.com/leaderboard/swe_bench_pro_public", "sources": [ "https://labs.scale.com/leaderboard/swe_bench_pro_public", "https://www.heyuan110.com/posts/ai/2026-07-10-gpt-5-6-general-availability/", "https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B", "https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B/blob/main/LICENSE", "https://arxiv.org/abs/2509.16941", "https://www.benchlm.ai/benchmarks/swePro" ] }, { "id": "swe-bench-verified", "name": "SWE-bench Verified", "version": null, "category": "coding", "measures": "Human-validated SWE-bench subset; near saturation at the top, contamination and test-design issues noted", "maintainer": "SWE-bench team (Princeton and Stanford), with OpenAI Preparedness for the Verified subset", "status": "contaminated", "url": "https://www.swebench.com/", "sources": [ "https://leaderboard.steel.dev/leaderboards/swe-bench-verified.md", "https://www.swebench.com/", "https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro", "https://openai.com/index/why-we-no-longer-evaluate-swe-bench-verified/", "https://benchlm.ai/benchmarks/swe-bench-verified", "https://github.com/SWE-bench/SWE-bench", "https://gigazine.net/gsc_news/en/20260429-swe-bench-verified" ] }, { "id": "swe-lancer-ic-diamond", "name": "SWE-Lancer IC Diamond", "version": null, "category": "coding", "measures": "Freelance software tasks", "maintainer": "openai", "status": "active", "url": null, "sources": [ "https://openai.com/index/introducing-gpt-5-3-codex" ] }, { "id": "swe-marathon", "name": "SWE-Marathon", "version": null, "category": "coding", "measures": "Long-horizon software engineering", "maintainer": null, "status": "active", "url": null, "sources": [ "https://huggingface.co/moonshotai/Kimi-K3" ] }, { "id": "tabarena", "name": "TabArena", "version": "All Tasks", "category": "tabular", "measures": "Elo across tabular tasks, anchored RandomForest default = 1000; TabPFN-lab researchers are involved", "maintainer": "TabArena team", "status": "active", "url": "https://benchmarklist.com/benchmarks/tabarena_all/", "sources": [ "https://benchmarklist.com/benchmarks/tabarena_all/", "https://mindfulmodeler.substack.com/p/the-state-of-tabular-foundation-models", "https://benchmarklist.com/benchmarks/tabarena_all/", "https://tabarena.ai/" ] }, { "id": "talent", "name": "TALENT", "version": null, "category": "tabular", "measures": "Tabular benchmark suggested as an independent cross-check to TabArena", "maintainer": null, "status": "active", "url": null, "sources": [ "https://mindfulmodeler.substack.com/p/the-state-of-tabular-foundation-models" ] }, { "id": "tar", "name": "TAR", "version": null, "category": "world models and robotics", "measures": "Vision for physical AI (cited by NVIDIA for Cosmos 3)", "maintainer": null, "status": "active", "url": null, "sources": [ "https://investor.nvidia.com/news/press-release-details/2026/NVIDIA-Launches-Cosmos-3-the-Open-Frontier-Foundation-Model-for-Physical-AI/default.aspx" ] }, { "id": "tau-bench", "name": "tau-bench", "version": null, "category": "agents and tool use", "measures": "Tool-using agent talks to a simulated user in airline, retail and telecom domains", "maintainer": "sierra", "status": "crowded", "url": "https://arxiv.org/abs/2406.12045", "sources": [ "https://arxiv.org/abs/2406.12045" ], "unverified": [ "maintainer" ] }, { "id": "tau2-bench", "name": "tau2-bench", "version": null, "category": "agents and tool use", "measures": "Second version of tau-bench; rows mix domains and pass^k settings", "maintainer": "sierra", "status": "crowded", "url": "https://www.benchlm.ai/benchmarks/tau2-bench", "sources": [ "https://www.benchlm.ai/benchmarks/tau2-bench" ], "unverified": [ "maintainer" ] }, { "id": "tau3-bench", "name": "tau3-bench", "version": null, "category": "agents and tool use", "measures": "Newer tau-bench release listed on BenchLM", "maintainer": "sierra", "status": "active", "url": null, "sources": [ "https://benchlm.ai/benchmarks" ], "unverified": [ "status" ] }, { "id": "teleqna", "name": "TeleQnA", "version": null, "category": "domain", "measures": "Telecom knowledge, 10,000 Q&A (arXiv 23 Oct 2023)", "maintainer": null, "status": null, "url": "https://arxiv.org/abs/2310.15051", "sources": [ "https://arxiv.org/abs/2310.15051" ] }, { "id": "tempcompass", "name": "TempCompass", "version": null, "category": "multimodal", "measures": "Temporal video understanding QA", "maintainer": null, "status": "active", "url": null, "sources": [ "https://arxiv.org/abs/2506.09985" ] }, { "id": "terminal-bench-2-0", "name": "Terminal-Bench", "version": "2.0", "category": "agents and tool use", "measures": "Agentic terminal tasks; frontier vendors have largely stopped reporting it", "maintainer": "Terminal-Bench team (Stanford, Harbor and Laude Institute), USA", "status": "superseded", "url": "https://www.tbench.ai/leaderboard", "sources": [ "https://rits.shanghai.nyu.edu/ai/openai-releases-gpt-5-5-agentic-coding-ceiling-tops-14-benchmarks/", "https://openai.com/index/introducing-gpt-5-3-codex", "https://arxiv.org/abs/2601.11868", "https://www.benchlm.ai/benchmarks/terminal-bench-2", "https://www.tbench.ai/leaderboard" ], "unverified": [ "maintainer" ] }, { "id": "terminal-bench-2-1", "name": "Terminal-Bench", "version": "2.1", "category": "agents and tool use", "measures": "Agentic terminal tasks", "maintainer": "Terminal-Bench", "status": "active", "url": "https://www.tbench.ai/leaderboard", "sources": [ "https://vals.ai/models/anthropic_claude-fable-5-1", "https://huggingface.co/moonshotai/Kimi-K3" ] }, { "id": "terminal-bench-3-0", "name": "Terminal-Bench", "version": "3.0", "category": "agents and tool use", "measures": "Agentic terminal tasks", "maintainer": "Terminal-Bench", "status": "active", "url": "https://www.tbench.ai/leaderboard", "sources": [ "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash", "https://huggingface.co/zai-org/GLM-5.3" ] }, { "id": "terminal-bench-4-0", "name": "Terminal-Bench", "version": "4.0", "category": "agents and tool use", "measures": "Agentic terminal tasks", "maintainer": "Terminal-Bench team (Stanford, Harbor and Laude Institute), USA", "status": "active", "url": "https://www.tbench.ai/leaderboard", "sources": [ "https://vpsranking.com/news/ai/ai-2026-09-22-anthropic-claude-opus-5-5/", "https://slash-digital.io/en/insights/ai-models-2026/", "https://llm-stats.com/blog/research/claude-opus-5-5-launch", "https://www.tbench.ai/leaderboard", "https://www.benchlm.ai/benchmarks/terminal-bench-2", "https://artificialanalysis.ai/", "https://arxiv.org/abs/2601.11868" ] }, { "id": "time", "name": "TIME", "version": null, "category": "time series", "measures": "Time-series forecasting", "maintainer": null, "status": null, "url": null, "sources": [ "https://datanorth.ai/nl/nieuws/google-lanceert-timesfm-3" ] }, { "id": "toolathlon", "name": "Toolathlon", "version": null, "category": "agents and tool use", "measures": "Agent benchmark tracked by BenchLM", "maintainer": null, "status": null, "url": null, "sources": [ "https://www.benchlm.ai/benchmarks" ], "unverified": [] }, { "id": "tsb-ad", "name": "TSB-AD", "version": null, "category": "time series", "measures": "Time-series anomaly detection, 1,070 series from 40 datasets (VUS-PR); NeurIPS 2024; open to community submissions from 1 Apr 2026", "maintainer": "TheDatumOrg", "status": "active", "url": "https://github.com/TheDatumOrg/TSB-AD", "sources": [ "https://arxiv.org/pdf/2602.01359", "https://github.com/TheDatumOrg/TSB-AD" ] }, { "id": "tts-arena", "name": "TTS Arena", "version": null, "category": "speech", "measures": "Blind pairwise human preference votes on which of two anonymous text-to-speech outputs sounds more human", "maintainer": "TTS-AGI (Hugging Face organisation)", "status": null, "url": "https://huggingface.co/spaces/TTS-AGI/TTS-Arena-V2", "sources": [ "https://tts-agi-tts-arena-v2.hf.space/about", "https://huggingface.co/spaces/TTS-AGI/TTS-Arena-V2" ] }, { "id": "usmle", "name": "USMLE", "version": null, "category": "domain", "measures": "US medical licensing exam questions", "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/m42-health/Llama3-Med42-70B" ] }, { "id": "vals-index", "name": "Vals Index", "version": null, "category": "composite index", "measures": "Composite of Vals.ai evaluations", "maintainer": "Vals.ai", "status": "active", "url": null, "sources": [ "https://www.orcarouter.ai/es/blog/gemini-4-argon" ] }, { "id": "vantage-bench", "name": "VANTAGE-Bench", "version": null, "category": "world models and robotics", "measures": "Vision for physical AI (cited by NVIDIA for Cosmos 3)", "maintainer": null, "status": "active", "url": null, "sources": [ "https://investor.nvidia.com/news/press-release-details/2026/NVIDIA-Launches-Cosmos-3-the-Open-Frontier-Foundation-Model-for-Physical-AI/default.aspx" ] }, { "id": "vbench", "name": "VBench", "version": "v1", "category": "generative media", "measures": "Submission-based text-to-video evaluation; Total and Quality scores", "maintainer": "Vchitect", "status": "active", "url": "https://huggingface.co/spaces/Vchitect/VBench_Leaderboard", "sources": [ "https://huggingface.co/spaces/Vchitect/VBench_Leaderboard" ] }, { "id": "vbench-2-0", "name": "VBench-2.0", "version": "2.0", "category": "generative media", "measures": "Submission-based video evaluation; Total and Creativity scores", "maintainer": "Vchitect", "status": "active", "url": "https://huggingface.co/spaces/Vchitect/VBench_Leaderboard", "sources": [ "https://huggingface.co/spaces/Vchitect/VBench_Leaderboard" ] }, { "id": "vending-bench-2", "name": "Vending-Bench 2", "version": null, "category": "agents and tool use", "measures": "Agent benchmark tracked by BenchLM", "maintainer": null, "status": null, "url": null, "sources": [ "https://www.benchlm.ai/benchmarks" ], "unverified": [] }, { "id": "vicuna-benchmark", "name": "Vicuna benchmark", "version": null, "category": "human preference arena", "measures": "Chatbot quality relative to ChatGPT", "maintainer": null, "status": "active", "url": null, "sources": [ "https://arxiv.org/abs/2305.14314" ] }, { "id": "video-mme", "name": "Video-MME", "version": null, "category": "multimodal", "measures": "Video understanding across 6 domains, clips from 11 s to 1 h, with subtitles and audio", "maintainer": "Video-MME team (academic, China)", "status": "active", "url": "https://video-mme.github.io/home_page.html", "sources": [ "https://video-mme.github.io/home_page.html" ], "unverified": [] }, { "id": "virchow-pan-cancer-detection", "name": "Pan-cancer specimen-level detection (Virchow study)", "version": null, "category": "domain", "measures": "Specimen-level pan-cancer detection AUC on pathology slides", "maintainer": null, "status": null, "url": null, "sources": [ "https://www.nature.com/articles/s41591-024-03141-0" ] }, { "id": "visa", "name": "VisA", "version": null, "category": "industrial", "measures": "Industrial image anomaly detection; near-saturated", "maintainer": null, "status": "crowded", "url": null, "sources": [ "https://arxiv.org/pdf/2510.17611" ] }, { "id": "visual-wake-words", "name": "Visual Wake Words (VWW)", "version": null, "category": "domain", "measures": "Person-presence image classification for microcontrollers (MCUNet repo)", "maintainer": null, "status": null, "url": null, "sources": [ "https://github.com/mit-han-lab/mcunet" ], "unverified": [] }, { "id": "vlabench", "name": "VLABench", "version": null, "category": "world models and robotics", "measures": "VLA benchmark", "maintainer": null, "status": "active", "url": null, "sources": [ "https://www.therobotreport.com/agibot-releases-go-2-foundation-model-embodied-ai/" ] }, { "id": "vlguard", "name": "VLGuard", "version": null, "category": "safety", "measures": null, "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/mistralai/Shieldstral-1.0-3B" ] }, { "id": "voxceleb1", "name": "VoxCeleb1 (cleaned)", "version": null, "category": "speech", "measures": "Speaker verification EER", "maintainer": null, "status": "static", "url": null, "sources": [ "https://www.pyannote.ai/blog/top-7-speaker-diarization-apis-and-models-in-2026" ] }, { "id": "weatherbench-2", "name": "WeatherBench 2", "version": "2", "category": "domain", "measures": "Global forecast skill (RMSE, CRPS, ACC); successor framework WeatherBench-X", "maintainer": "google", "status": "active", "url": "https://sites.research.google/gr/weatherbench/", "sources": [ "https://sites.research.google/gr/weatherbench/", "https://researchsites.withgoogle.com/weatherbench/deterministic/", "https://developers.google.com/weathernext/guides/evals" ] }, { "id": "webarena", "name": "WebArena", "version": null, "category": "agents and tool use", "measures": "Agent benchmark tracked by BenchLM", "maintainer": null, "status": null, "url": null, "sources": [ "https://www.benchlm.ai/benchmarks" ], "unverified": [] }, { "id": "webarena-verified", "name": "WebArena-Verified", "version": null, "category": "agents and tool use", "measures": "Agent benchmark tracked by BenchLM", "maintainer": null, "status": null, "url": null, "sources": [ "https://www.benchlm.ai/benchmarks" ], "unverified": [] }, { "id": "wildguardtest", "name": "WildGuardTest", "version": null, "category": "safety", "measures": null, "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/mistralai/Shieldstral-1.0-3B", "https://huggingface.co/nvidia/Nemotron-3.5-Content-Safety" ] }, { "id": "worldscore", "name": "WorldScore", "version": null, "category": "world models and robotics", "measures": "Controllability, quality and dynamics of world generation; Static and Dynamic aggregates", "maintainer": "Duan et al., Stanford (ICCV 2025)", "status": "active", "url": "https://arxiv.org/html/2504.00983v2", "sources": [ "https://arxiv.org/html/2504.00983v2" ], "unverified": [ "measures", "status" ] }, { "id": "xstest", "name": "XSTest", "version": null, "category": "safety", "measures": null, "maintainer": null, "status": null, "url": null, "sources": [ "https://huggingface.co/mistralai/Shieldstral-1.0-3B" ] } ]