Models
Models from the labs in Epoch AI's catalog, the repositories they published on Hugging Face, and posts from their own news pages.
GPT-6.1 TerraOpenAI · Language, Multimodal, VisionAPI accessOct 26DeepSeek V4.1 Flash 1017DeepSeek · Language, Multimodal, VisionOpen weights (unrestricted)Oct 17MAI-Base-2Microsoft · Language, Vision, Multimodal · 2,800B parameters · Training compute 2.4e25 FLOPUnreleasedOct 15Muse Spark OpenMeta AI · Language, Multimodal, Vision, Audio, VideoOpen weights (restricted use)Oct 15Muse WonderMeta AI · Language, Vision, Video, MultimodalAPI accessOct 12Muse Wonder FullMeta AI · Language, Vision, Video, MultimodalAPI accessOct 12Claude Haiku 5.5Anthropic · Language, VisionAPI accessOct 7Gemini 4 ArgonGoogle DeepMind · Language, Multimodal, VisionLimited accessSep 30GPT-6.1 SolOpenAI · Multimodal, Language, VisionAPI accessSep 29Claude Sonnet 5.5Anthropic · Language, Multimodal, VisionAPI accessSep 28Claude Opus 5.5Anthropic · LanguageAPI accessSep 22GPT-6 LunaOpenAI · Multimodal, Language, VisionAPI accessSep 22GPT-6 Luna DecisionsOpenAI · Language, Vision, MultimodalAPI accessSep 22GPT-6 SolOpenAI · Multimodal, Language, VisionAPI accessSep 22Grok 4.7xAI · Language · 2.1B parametersAPI accessSep 21DeepSeek V4.1 FlashDeepSeek · Language, Vision, Multimodal · 748B parameters · Training compute 4.3e24 FLOPOpen weights (unrestricted)Sep 9ChatGPT Images 2.5OpenAI · Image generation, VisionHosted access (no API)Sep 8GPT-Image-2.5 FlareOpenAI · Image generationAPI accessSep 8GPT-Image-2.5 SunburstOpenAI · Image generationAPI accessSep 8
GPT-6 AstraOpenAI · Multimodal, Language, Vision · Training compute 9.4e26 FLOPAPI accessSep 3
WeatherNext 3Google DeepMind, Google Research · Earth science · Training compute 2.1e23 FLOPHosted access (no API)Sep 3Gemini 3.8 FlashGoogle DeepMind · Language, Multimodal, Image generation, Video, AudioAPI accessSep 2Muse Spark 1.3Meta AI · LanguageAPI accessSep 2Claude Fable 5.1Anthropic · Language, Multimodal, VisionAPI accessSep 1GLM-5.3-FlashZ.ai (Zhipu AI) · Language, Vision, Multimodal · 320B parameters · Training compute 3.2e24 FLOPOpen weights (unrestricted)Aug 20Data: Epoch AI, CC BY 4.0. Changes: selected rows and columns of Epoch AI's database.