Project AURIONAI Use-Cases Research & Innovation
OpenAI

OpenAI

Frontier lab, USA. 100 models in the directory.

frontier labUSA
0deployments

Identity

Type
frontier lab
Country label
USA
HQ country
USA
Founding country
USA
Parent
not in the notes
Parent country
not in the notes
Country basis
the research notes
Where this comes from
Convention D1: a dual label, HQ first, when the HQ, founding or parent country differ.

Models by Class

LLM29

Models

ModelClassOpennessParametersReleased
OpenAIGPT-6.1 Solmultimodal LLMproprietary–29 Sep 2026OpenAIGPT-6.1 AstraLLMproprietary–28 Sep 2026OpenAIGPT-6 Lunamultimodal LLMproprietary–22 Sep 2026OpenAIGPT-6 Solmultimodal LLMproprietary–22 Sep 2026OpenAIGPT Image 2.5 Flareimage generationproprietary–8 Sep 2026OpenAIGPT Image 2.5 Sunburstimage generationproprietary–8 Sep 2026OpenAIGPT-6 Astramultimodal LLMproprietary–3 Sep 2026OpenAIGPT-5.6 LunaLLMproprietary–9 Jul 2026OpenAIGPT-5.6 SolLLMproprietary–9 Jul 2026OpenAIGPT-5.6 TerraLLMproprietary–9 Jul 2026OpenAIGPT-5.5LLMproprietary–23 Apr 2026OpenAIOpenAI privacy-filterguardunknown1B22 Apr 2026OpenAIGPT Image 2image generationproprietary–Apr 2026OpenAIGPT-5.3-Codexcodeproprietary–5 Feb 2026OpenAIgpt-oss-safeguard-20bguardopen weights permissive21BJan 2026OpenAIGPT-5.1-Codex-Maxcodeproprietary–4 Dec 2025OpenAIGPT Image 1.5image generationproprietary–Dec 2025OpenAIGPT-5.1LLMproprietary–13 Nov 2025OpenAIGPT-5.1-Codexcodeproprietary–13 Nov 2025OpenAIGPT-5.1-Codex-Minicodeproprietary–13 Nov 2025OpenAIGPT-5 proreasoningproprietary–6 Oct 2025OpenAIgpt-oss-safeguard-120bguardopen weights permissive117BOct 2025OpenAIGPT-5LLMproprietary–7 Aug 2025OpenAIGPT-5 nanoreasoningproprietary–7 Aug 2025OpenAIgpt-ossLLMopen weights permissive–5 Aug 2025OpenAIgpt-oss-120bLLMopen weights permissive117B5 Aug 2025OpenAIgpt-oss-20bLLMopen weights permissive21B5 Aug 2025OpenAIo3-proreasoningproprietary–10 Jun 2025OpenAIo3reasoningproprietary–16 Apr 2025OpenAIo4-minireasoningproprietary–16 Apr 2025OpenAIGPT-4.1LLMproprietary–14 Apr 2025OpenAIGPT-4.1multimodal LLMproprietary–14 Apr 2025OpenAIGPT-4.1 miniLLMproprietary–14 Apr 2025OpenAIGPT-4.1 nanoLLMproprietary–14 Apr 2025OpenAIGPT-4o (March 2025, chatgpt-4o-latest)multimodal LLMproprietary–27 Mar 2025OpenAIGPT-4o (March 2025 update)multimodal LLMproprietary–27 Mar 2025OpenAIo1-proreasoningproprietary–19 Mar 2025OpenAIcomputer-use-previewmultimodal LLMproprietary–11 Mar 2025OpenAIo3-minireasoningproprietary–31 Jan 2025OpenAIo1reasoningunknown–12 Sep 2024OpenAIo1-minireasoningproprietary–12 Sep 2024OpenAIGPT-4o miniLLMproprietary–18 Jul 2024OpenAIWhisper large-v3-turbospeech recognitionopen weights permissive809M2024OpenAIDALL-E 3image generationproprietary–Sep 2023OpenAIShap-E3Dopen weights permissive–3 May 2023OpenAIWhisper large-v3speech recognitionopen weights permissive1.55B2023OpenAIwhisper-large-v2speech recognitionopen weights permissive1.54B5 Dec 2022OpenAIwhisper-smallspeech recognitionopen weights permissive–26 Sep 2022OpenAIwhisper-basespeech recognitionopen weights permissive72.59M26 Sep 2022OpenAIwhisper-mediumspeech recognitionopen weights permissive763.86M26 Sep 2022OpenAICLIP ViT-L/14 336pxembeddingunknown–22 Apr 2022OpenAIDALL-E 2image generationproprietary–13 Apr 2022OpenAIGLIDEimage generationunknown–20 Dec 2021OpenAIGuided Diffusion (ADM)image generationunknown–11 May 2021OpenAICLIPembeddingunknown–26 Feb 2021OpenAIImproved DDPMimage generationunknown–18 Feb 2021OpenAIOpenAI Fivereinforcement learningunknown–13 Dec 2019OpenAIGPT-3 175BLLMunknown175B–OpenAIGPT-3.5 Turbo (0613)LLMproprietary––OpenAIGPT-4LLMproprietary––OpenAIgpt-4o-mini-ttsspeech synthesisproprietary––OpenAIGPT-5.2LLMproprietary––OpenAIGPT-5.4LLMproprietary––OpenAIGPT-5.6 CyberLLMproprietary––OpenAIGPT-5 miniLLMproprietary––OpenAIGPT-6LLMproprietary––OpenAIGPT Imageimage generationproprietary––OpenAIgpt-live-1speech synthesisproprietary––OpenAIgpt-realtime-1.5speech synthesisproprietary––OpenAIgpt-realtime-2speech synthesisproprietary––OpenAIgpt-realtime-2.1speech synthesisproprietary––OpenAIgpt-realtime-2.1-minispeech synthesisproprietary––OpenAIgpt-realtime-minispeech synthesisproprietary––OpenAIgpt-realtime-translatespeech synthesisproprietary––OpenAIo-seriesLLMproprietary––OpenAIOpenAI Moderation (omni-moderation-latest)guardproprietary––OpenAISoravideo generationproprietary––OpenAISora 2video generationproprietary––OpenAISora 2 Provideo generationproprietary––OpenAISoravideo generationunknown––OpenAIWhisperspeech recognitionopen weights permissive––OpenAIwhisper-tinyspeech recognitionunknown––OpenAIGPT-2LLMopen weights permissive––OpenAIGPT-2 (124M)LLMopen weights permissive137.02M–OpenAIGPT-2 mediumLLMopen weights permissive379.99M–OpenAIGPT-2 largeLLMopen weights permissive811.78M–OpenAIGPT-2 XLLLMopen weights permissive1.61B–OpenAIGPT (GPT-1)LLMopen weights permissive119.68M–OpenAICLIP ViT-B/32embeddingunknown––OpenAICLIP ViT-L/14embeddingunknown427.62M–OpenAIOpenAI text embeddingsembeddingproprietary––OpenAItext-embedding-3-smallembeddingproprietary––OpenAItext-embedding-3-largeembeddingproprietary––OpenAItext-embedding-ada-002embeddingproprietary––OpenAIGPT-5.2-Codexcodeproprietary––OpenAIcodex-mini-latestcodeproprietary––OpenAIgpt-4o-transcribespeech recognitionproprietary––OpenAIDALL-Eimage generationunknown––OpenAIGPT Image 1image generationproprietary––OpenAIGPT Image 1 miniimage generationproprietary––

Scores

BenchmarkModelScoreWho ran itSetting
ARC-AGI-2GPT-6 Astra95 %vendor-reported$1.12/task
ARC-AGI-3GPT-6 Astrarank onlyvendor-reporteddirect, about 63 to 66% (apiyi citing Latent Space)
ARC-AGI-3GPT-6 Astra98.6 %vendor-reportedConflicting figures across sources
ARC-AGI-3GPT-6 Astra99.9 %vendor-reportedwith OpenAI adapter harness (apiyi citing Latent Space)
Arena Text leaderboardGPT-5.6 Sol1,485 Eloindependentxhigh
Arena Text leaderboardGPT-6 Astra1,475 Eloindependentmax
Arena Text leaderboardGPT-6.1 Sol1,484 Eloindependentmax
Arena Vision leaderboardGPT-6 Astra1,286 Eloindependentmax
Arena Vision leaderboardGPT-6.1 Sol1,294 Eloindependentmax
Artificial Analysis Intelligence IndexGPT-5.457 indexindependentGPT-5.4 (xhigh); v4.0, stale; superseded by AA v4.3.2
Artificial Analysis Intelligence IndexGPT-6 Astra53 indexindependentmax; 52.67 per secondary
Artificial Analysis Intelligence IndexGPT-6 Luna38 indexindependentmax
Artificial Analysis Intelligence IndexGPT-6.1 Sol52 indexindependentmax
Artificial Analysis Text-to-Image ArenaGPT Image 1.51,107 Eloindependent(high); rank shown as 9-14
Artificial Analysis Text-to-Image ArenaGPT Image 21,171 Eloindependent(high)
Artificial Analysis Text-to-Image ArenaGPT Image 2.5 Flare1,191 Eloindependent(max)
Artificial Analysis Text-to-Image ArenaGPT Image 2.5 Sunburst1,198 Eloindependent(max)
BenchLM long-context compositeGPT-5.585.3 indexvendor-reportedComposite of vendor-reported scores
benchlm.ai frontier compositeGPT-6 Astra85 indexvendor-reported
benchlm.ai frontier compositeGPT-6.1 Sol81.5 indexvendor-reportedestimated
ConvFinQAGPT-476.48 scoreindependentfew-shot
DeepSWEGPT-6 Astra74.1 %vendor-reported
DeepSWE (version not stated)GPT-5.6 Sol73 %vendor-reported
Epoch Capabilities Index (ECI)GPT-6 Astra166.41 indexmaintainer-runEpoch data file of 15 Oct 2026: 0.41 behind Claude Opus 5.5, within each other's confidence intervals
FinanceBenchGPT-5.591 %vendor-reportedAggregator score, not comparable to paper setup; not independently confirmed
FiQA sentimentGPT-488.11 scoreindependentfew-shot
FrontierMathGPT-5.6 Sol89 %vendor-reportedLegacy aggregate mixing benchmark versions; self-report
FrontierMath Tier 4GPT-6 Astra97.6 %vendor-reported
FrontierMath: ErdosGPT-6 Astra3 %maintainer-runSolved 2 of 68 problems
GAIAGPT-415 %maintainer-runGPT-4 with plugins, at benchmark release
GPQA DiamondGPT-5.6 Sol94.6 %vendor-reported
GPQA DiamondGPT-6 Astra96 %vendor-reported
GPQA DiamondGPT-6 Astra96 %vendor-reported
HealthBench HardGPT-546.2 %vendor-reported
Hugging Face Open ASR LeaderboardWhisper large-v37.4 WERmaintainer-runaverage WER per Gladia, from the same leaderboard
Hugging Face Open ASR LeaderboardWhisper large-v37.44 WERmaintainer-runleadership history
Hugging Face Open ASR LeaderboardWhisper large-v3-turbo7.75 WERmaintainer-runaverage WER per Gladia, from the same leaderboard
IDSV (Indonesian Speech Variability)whisper-small13.84 CERindependentIDSV test, all, zero-shot
IDSV (Indonesian Speech Variability)whisper-small30.87 WERindependentIDSV test, all, zero-shot
LMArena Text-to-Image ArenaGPT Image 21,383 Eloindependentmedium
LMArena Text-to-Image ArenaGPT Image 2.5 Flare1,398 Eloindependentpreliminary
LMArena Text-to-Image ArenaGPT Image 2.5 Sunburst1,425 Eloindependentpreliminary
LMArena Text-to-Video ArenaSora 2 Pro1,368 Eloindependent
OpenAI MRCRGPT-6 Astra96.3 %vendor-reported512K to 1M
OSWorldGPT-6 Astra72.6 %vendor-reported
OSWorld-VerifiedGPT-5.3-Codex64.7 %vendor-reportedxhigh
Qwen-Image-BenchGPT Image 2.5 Sunburst67.01 scorevendor-reportedQwen chart only; methodology unreleased
SEA-HELMgpt-oss-120b66.14 scoremaintainer-run
SEA-HELMgpt-oss-120b72.28 scoremaintainer-run
SEA-HELMgpt-oss-20b60.64 scoremaintainer-runCheckpoint gpt-oss-20b as listed, quantization mxfp4; release view (data last updated 18 Sep 2026); score is the average of 2000 bootstraps
SEA-HELMgpt-oss-20b68.38 scoremaintainer-runCheckpoint gpt-oss-20b as listed, quantization mxfp4; release view (data last updated 18 Sep 2026); score is the average of 2000 bootstraps
SWE-bench ProGPT-5.3-Codex56.8 %vendor-reportedxhigh
SWE-bench ProGPT-5.558.6 %vendor-reported
SWE-bench ProGPT-5.6 Sol64.6 %vendor-reportedpublic board (per heyuan110)
SWE-bench VerifiedGPT-5.6 Sol82.2 %independentmax/xhigh, third-party via Thinking Machines
SWE-Lancer IC DiamondGPT-5.3-Codex81.4 %vendor-reportedxhigh
Terminal-BenchGPT-5.3-Codex77.3 %vendor-reportedxhigh
Terminal-BenchGPT-5.582.7 %vendor-reported
Terminal-BenchGPT-5.582 %vendor-reported
Terminal-BenchGPT-6 Astra57.9 %vendor-reported57.9% to 58.2% across sources
VBench-2.0Sora58.38 %maintainer-runTotal score; Sora-480p
VBench-2.0Sora60.57 %maintainer-runCreativity score; Sora-480p
Where this comes from
Every recorded score of this organisation's models, with who ran it. Scores from different settings are not directly comparable.

Sources

  1. Models | OpenAI API developers.openai.com https://developers.openai.com/api/docs/models
  2. LLM Leaderboard - Comparison of AI models from OpenAI, Anthropic, Google, SpaceXAI & others | Artificial Analysis artificialanalysis.ai https://artificialanalysis.ai/leaderboards/models
  3. openai/gpt-oss-120b , Hugging Face huggingface.co https://huggingface.co/openai/gpt-oss-120b
  4. huggingface.co/openai https://huggingface.co/openai
  5. openai/gpt-oss-safeguard-20b , Hugging Face huggingface.co https://huggingface.co/openai/gpt-oss-safeguard-20b
  6. Gladia - Best open-source speech-to-text models in 2026 gladia.io https://www.gladia.io/blog/best-open-source-speech-to-text-models
  7. OpenAI - Wikipedia en.wikipedia.org https://en.wikipedia.org/wiki/OpenAI
  8. BrowseComp: A Simple Yet Challenging Benchmark for Browsing Agents arxiv.org https://arxiv.org/abs/2504.12516
  9. openai.com/index/why-we-no-longer-evaluate-swe-bench-ver... https://openai.com/index/why-we-no-longer-evaluate-swe-bench-verified/
  10. openai.com/index/openai-for-healthcare https://openai.com/index/openai-for-healthcare/
  11. openai.com/index/introducing-gpt-5 https://openai.com/index/introducing-gpt-5/
  12. Training compute of frontier AI models grows by 4-5x per year epoch.ai https://epoch.ai/blog/training-compute-of-frontier-ai-models-grows-by-4-5x-per-year
  13. Over 30 AI models have been trained at the scale of GPT-4 epoch.ai https://epoch.ai/data-insights/models-over-1e25-flop
  14. Learning to reason with LLMs openai.com https://openai.com/index/learning-to-reason-with-llms/
  15. openai.com/index/introducing-gpt-oss https://openai.com/index/introducing-gpt-oss/

To cite this page: Project AURION, AI Use-Cases Research & Innovation, "OpenAI", https://aurion.semai.id/organisations/openai/, records of October 2026.