Skip to content
fedi.software

AI models of 2025: who released what

Every notable model release month by month: vendor and its country, open or closed weights, type, context window, size of open models — and which tasks the models of each year lead today.

2025 155 releases · 69 with open weights

Models of 2025 leading each task today

December

  1. A MiniMax M2.1 HFMiniMaxChinaOpen weightsLLM205K229B / 11B
  2. A GLM 4.7 HFZhipu AIChinaOpen weightsLLM205K358B / 39.2B
  3. S Gemini 3 Flash PreviewGoogleUnited StatesClosedVision (VLM)1M
  4. B flux-2-maxBlack Forest LabsGermanyClosedImage
  5. A MiMo V2 FlashXiaomiChinaOpen weightsLLM262K
  6. B seedance-v1.5-proByteDanceChinaClosedVideo
  7. B Nemotron 3 Nano 30B A3B HFNVIDIAUnited StatesOpen weightsLLM131K31.6B / 3.5B
  8. A GPT 5.2OpenAIUnited StatesClosedVision (VLM)128K
  9. C GPT 5.2 CodexOpenAIUnited StatesClosedVision (VLM)400K
  10. C mixtral-8x22b-instruct-v0.1Mistral AIFranceOpen weightsLLM66K
  11. Devstral 2 2512 HFMistral AIFranceOpen weightsLLM262K125B
  12. B GLM 4.6V HFZhipu AIChinaOpen weightsVision (VLM)128K108B / 14.3B
  13. B seedream-4.5ByteDanceChinaClosedImage
  14. Ministral 3 14B 2512 HFMistral AIFranceOpen weightsVision (VLM)262K13.9B
  15. A mistral-large-3Mistral AIFranceClosedVision (VLM)262K
  16. B Nova 2 LiteAmazonUnited StatesClosedVision (VLM)1M
  17. A DeepSeek V3.2 HFDeepSeekChinaClosedLLM164K
  18. Ministral 3 3B 2512 HFMistral AIFranceOpen weightsVision (VLM)131K3.8B
  19. C Ministral 3 8B 2512 HFMistral AIFranceOpen weightsVision (VLM)262K8.9B
  20. B qwen-image-2512AlibabaChinaClosedImage

November

  1. B flux-2-flexBlack Forest LabsGermanyClosedImage
  2. B flux-2-proBlack Forest LabsGermanyClosedImage
  3. A gpt-image-1.5OpenAIUnited StatesClosedVision (VLM)128K
  4. S Claude Opus 4.5AnthropicUnited StatesClosedVision (VLM)200K
  5. A Nano Banana Pro (Gemini 3 Pro Image Preview)GoogleUnited StatesClosedVision (VLM)131K
  6. S gemini-3-proGoogleUnited StatesClosedVision (VLM)1M
  7. A grok-4.1-thinkingxAIUnited StatesClosedVision (VLM)200K
  8. A GPT 5.1OpenAIUnited StatesClosedVision (VLM)400K
  9. C GPT 5.1 CodexOpenAIUnited StatesClosedVision (VLM)400K
  10. C GPT 5.1 Codex MiniOpenAIUnited StatesClosedLLM400K
  11. Kimi K2 Thinking HFMoonshot AIChinaOpen weightsReasoning262K1,026B / 32B
  12. A kimi-k2-thinking-turboMoonshot AIChinaClosedLLM262K
  13. A longcat-flash-chat-2602-expMeituanChinaClosedLLM131K

October

  1. B MiniMax M2 HFMiniMaxChinaOpen weightsLLM205K229B / 11B
  2. Qwen3 VL 32B Instruct HFAlibabaChinaClosedVision (VLM)262K
  3. Granite 4.0 Micro HFIBMUnited StatesOpen weightsLLM131K3.2B
  4. A Claude Haiku 4.5AnthropicUnited StatesClosedVision (VLM)200K
  5. A Veo 3.1GoogleUnited StatesClosedVideo
  6. A Veo 3.1 fastGoogleUnited StatesClosedVideo
  7. Qwen3 VL 8B Instruct HFAlibabaChinaClosedVision (VLM)262K
  8. Qwen3 VL 8B Thinking HFAlibabaChinaClosedReasoning131K
  9. Qwen3 VL 30B A3B Thinking HFAlibabaChinaClosedReasoning262K
  10. B sora-2OpenAIUnited StatesClosedVideo
  11. A sora-2-proOpenAIUnited StatesClosedVideo

September

  1. A GLM 4.6 HFZhipu AIChinaOpen weightsLLM205K357B / 37.6B
  2. S Claude Sonnet 4.5AnthropicUnited StatesClosedVision (VLM)1M
  3. A DeepSeek V3.2 Exp HFDeepSeekChinaOpen weightsLLM164K685B / 51.9B
  4. B gemini-2.5-flash-lite-preview-09-2025-no-thinkingGoogleUnited StatesClosedVision (VLM)1M
  5. A gemini-2.5-flash-preview-09-2025GoogleUnited StatesClosedVision (VLM)1M
  6. B gpt-image-1-miniOpenAIUnited StatesClosedVision (VLM)
  7. A Qwen3 MaxAlibabaChinaClosedLLM262K
  8. A Qwen3 VL 235B A22B Instruct HFAlibabaChinaOpen weightsVision (VLM)131K236B / 22.8B
  9. A Qwen3 VL 235B A22B Thinking HFAlibabaChinaOpen weightsReasoning131K236B / 22.8B
  10. A DeepSeek V3.1 Terminus HFDeepSeekChinaOpen weightsLLM128K685B / 51.1B
  11. Kimi K2 0905 HFMoonshot AIChinaClosedLLM262K
  12. D grok-code-fast-1xAIUnited StatesClosedLLM256K
  13. A Qwen3 Next 80B A3B Instruct HFAlibabaChinaOpen weightsLLM131K81.3B / 3B
  14. B Qwen3 Next 80B A3B Thinking HFAlibabaChinaOpen weightsReasoning131K81.3B / 3B

August

  1. Qwen3 30B A3B Thinking 2507 HFAlibabaChinaClosedReasoning82K
  2. Hermes 4 405B HFNous ResearchUnited StatesOpen weightsLLM128K406B
  3. B Nano Banana (Gemini 2.5 Flash Image)GoogleUnited StatesClosedVision (VLM)33K
  4. A DeepSeek V3.1 HFDeepSeekChinaOpen weightsLLM164K685B / 51.1B
  5. D flux-1-kontext-devBlack Forest LabsGermanyOpen weightsImage
  6. A Mistral Medium 3.1Mistral AIFranceClosedVision (VLM)262K
  7. B GLM 4.5V HFZhipu AIChinaOpen weightsVision (VLM)64K108B / 12B
  8. A GPT 5OpenAIUnited StatesClosedVision (VLM)400K
  9. B GPT-5 MiniOpenAIUnited StatesClosedVision (VLM)400K
  10. B GPT-5 NanoOpenAIUnited StatesClosedVision (VLM)400K
  11. S Claude Opus 4.1AnthropicUnited StatesClosedVision (VLM)197K
  12. B gpt-oss-120b HFOpenAIUnited StatesOpen weightsReasoning131K117B / 5.1B
  13. B gpt-oss-20b HFOpenAIUnited StatesOpen weightsReasoning131K20.9B / 3.6B
  14. Qwen3 4B Instruct 2507 HFAlibabaChinaOpen weightsLLM4B

July

  1. Qwen3 Coder 30B A3B Instruct FP8 HFAlibabaChinaOpen weightsLLM30.5B / 3.3B
  2. B step-3StepFunChinaClosedVision (VLM)66K
  3. B Qwen3 30B A3B Instruct 2507 HFAlibabaChinaOpen weightsLLM256K30.5B / 3.3B
  4. A GLM 4.5 HFZhipu AIChinaOpen weightsLLM131K358B / 39.2B
  5. B GLM 4.5 Air HFZhipu AIChinaOpen weightsLLM131K110B / 17B
  6. B Llama 3.3 Nemotron Super 49B v1.5NVIDIAUnited StatesOpen weightsLLM131K
  7. A Qwen3 235B A22B Thinking 2507 HFAlibabaChinaClosedReasoning131K
  8. B Qwen3 Coder 480B A35B HFAlibabaChinaOpen weightsLLM262K480B / 35B
  9. Qwen3 Coder 480B A35B Instruct FP8 HFAlibabaChinaOpen weightsLLM480B / 35B
  10. UI-TARS 7B HFByteDanceChinaClosedVision (VLM)128K
  11. Qwen3 235B A22B Instruct 2507 HFAlibabaChinaClosedLLM262K
  12. Voxtral Small 24B 2507 HFMistral AIFranceOpen weightsLLM33K24.3B
  13. A hunyuan-t1-20250711TencentChinaClosedLLM131K
  14. B Kimi K2 0711 HFMoonshot AIChinaOpen weightsLLM256K1,026B / 32B
  15. A grok-4-0709xAIUnited StatesClosedVision (VLM)256K
  16. Hunyuan A13B Instruct HFTencentChinaClosedLLM131K
  17. D Devstral MediumMistral AIFranceOpen weightsLLM262K

June

  1. ERNIE 4.5 VL 424B A47B HFBaiduChinaClosedVision (VLM)123K
  2. B Mistral Small 3.2Mistral AIFranceOpen weightsVision (VLM)128K
  3. Mistral Small 3.2 24B HFMistral AIFranceOpen weightsVision (VLM)128K24B
  4. A Gemini 2.5 FlashGoogleUnited StatesClosedVision (VLM)1M
  5. B Gemini 2.5 Flash LiteGoogleUnited StatesClosedVision (VLM)1M
  6. S Gemini 2.5 ProGoogleUnited StatesClosedVision (VLM)1M
  7. B MiniMax M1MiniMaxChinaOpen weightsLLM1M
  8. B gemma-3n-e4b-itGoogleUnited StatesOpen weightsVision (VLM)128K
  9. D lucid-originLeonardoAustraliaClosedImage

May

  1. DeepSeek R1 0528 Qwen3 8B HFDeepSeekChinaOpen weightsReasoning8.2B
  2. A DeepSeek R1 0528 HFDeepSeekChinaClosedReasoning164K
  3. D mistral-7b-instruct-v0.2Mistral AIFranceClosedLLM32K
  4. B Claude Sonnet 4AnthropicUnited StatesClosedVision (VLM)983K
  5. A claude-opus-4-20250514-thinking-16kAnthropicUnited StatesClosedVision (VLM)193K
  6. A veo-3-audioGoogleUnited StatesClosedVideo
  7. B Mistral Medium 3Mistral AIFranceClosedVision (VLM)131K

April

  1. B Qwen3 30B A3B HFAlibabaChinaOpen weightsReasoning41K30.5B / 3.3B
  2. Qwen3 1.7B Base HFAlibabaChinaOpen weightsLLM1.7B
  3. Qwen3 8B Base HFAlibabaChinaOpen weightsLLM8.2B
  4. Qwen3 1.7B HFAlibabaChinaOpen weightsLLM2B
  5. Qwen3 4B HFAlibabaChinaOpen weightsReasoning4B
  6. B gpt-image-1OpenAIUnited StatesClosedImage
  7. B hunyuan-turbos-20250416TencentChinaClosedLLM131K
  8. A OpenAI o3OpenAIUnited StatesClosedVision (VLM)200K
  9. B o4 MiniOpenAIUnited StatesClosedVision (VLM)200K
  10. B Llama 3.1 Nemotron 70B InstructNVIDIAUnited StatesOpen weightsLLM128K
  11. B GPT 4.1OpenAIUnited StatesClosedVision (VLM)1M
  12. B GPT-4.1 MiniOpenAIUnited StatesClosedVision (VLM)1M
  13. B GPT-4.1 NanoOpenAIUnited StatesClosedVision (VLM)1M
  14. B grok-3-mini-highxAIUnited StatesClosedLLM131K
  15. A grok-3-preview-02-24xAIUnited StatesClosedLLM131K
  16. B Llama 3.1 Nemotron Ultra 253BNVIDIAUnited StatesOpen weightsLLM128K
  17. B Llama 4 Maverick HFMetaUnited StatesOpen weightsVision (VLM)1M402B
  18. B Llama 4 Scout HFMetaUnited StatesOpen weightsVision (VLM)328K109B / 18B
  19. Qwen3 14B HFAlibabaChinaOpen weightsReasoning131K14.8B
  20. A Qwen3 235B A22B HFAlibabaChinaOpen weightsReasoning131K235B / 22B
  21. B Qwen3 32B HFAlibabaChinaOpen weightsReasoning131K32.8B
  22. Qwen3 8B HFAlibabaChinaOpen weightsReasoning131K8.2B
  23. Qwen3 Coder 30B A3B Instruct HFAlibabaChinaOpen weightsLLM262K30.5B / 3.3B
  24. Qwen3 VL 30B A3B Instruct HFAlibabaChinaOpen weightsVision (VLM)131K31.1B / 3.9B

March

  1. A chatgpt-4o-latest-20250326OpenAIUnited StatesClosedVision (VLM)128K
  2. DeepSeek V3 0324 HFDeepSeekChinaClosedLLM128K
  3. B Magistral Medium (latest)Mistral AIFranceOpen weightsLLM128K
  4. B Mistral Small 3.1 24B HFMistral AIFranceOpen weightsVision (VLM)128K24B
  5. B Command A HFCohereCanadaOpen weightsLLM256K111B
  6. B Gemma 3 12B HFGoogleUnited StatesOpen weightsVision (VLM)131K12.2B
  7. B Gemma 3 27B HFGoogleUnited StatesOpen weightsVision (VLM)131K27.4B
  8. B Gemma 3 4B HFGoogleUnited StatesOpen weightsVision (VLM)131K4.3B
  9. Reka Flash 3 HFReka AIUnited StatesClosedLLM66K
  10. B QwQ 32BAlibabaChinaOpen weightsLLM24K

February

  1. A gpt-4.5-preview-2025-02-27OpenAIUnited StatesClosedLLM128K
  2. D ray2Luma AIUnited StatesClosedVideo
  3. B claude-3-7-sonnet-20250219-thinking-32kAnthropicUnited StatesClosedVision (VLM)200K
  4. B glm-4-plus-0111Zhipu AIChinaClosedLLM128K
  5. B o3 Mini HighOpenAIUnited StatesClosedVision (VLM)200K
  6. B hunyuan-large-2025-02-10TencentChinaClosedLLM
  7. B hunyuan-standard-2025-02-10TencentChinaClosedLLM
  8. B gemini-2.0-flash-lite-preview-02-05GoogleUnited StatesClosedVision (VLM)990K

January

  1. B o3 MiniOpenAIUnited StatesClosedLLM200K
  2. C Mistral Small 3 HFMistral AIFranceOpen weightsLLM33K23.6B
  3. DeepSeek R1 Distill Llama 70B HFDeepSeekChinaOpen weightsReasoning70.6B
  4. DeepSeek R1 Distill Qwen 1.5B HFDeepSeekChinaOpen weightsReasoning1.8B
  5. DeepSeek R1 Distill Qwen 32B HFDeepSeekChinaOpen weightsReasoning32.8B
  6. B DeepSeek R1 HFDeepSeekChinaOpen weightsReasoning80K684B / 37B
  7. MiniMax 01 HFMiniMaxChinaClosedLLM1M
  8. C Phi 4 HFMicrosoftUnited StatesClosedLLM16K

Release dates come from the model catalogues (models.dev, LiteLLM, Hugging Face); models known only by the year of release are not shown. The leaders of a year are its models that rank highest on our leaderboards today.

Updated · Sources: models.dev (MIT), LiteLLM (MIT), LMArena (CC BY 4.0), Hugging Face

Arena (LMArena) scores: Arena Leaderboard Dataset by LMArena, CC BY 4.0. Modified by fedi.software (filtered, re-ranked, aggregated). Not endorsed by LMArena.

A release calendar of AI labs

This chart puts model releases on a single line of time, newest year first and grouped by month. Each line is one model with its vendor, the country where that company is based, whether the weights are open or closed, its type and its context window. For open models you also see the number of parameters; for a mixture-of-experts (MoE) model the second number is the active part used for each token.

Only models from AI labs make it in: those rated on our leaderboards or with weights published on Hugging Face. Resellers, routers and community fine-tunes are left out, so the list stays a record of who actually built what.

Reading the badges

  • Open weights — the files can be downloaded and run on your own machine; which model fits your hardware tells you what that takes.
  • Closed — the model is reachable only through the vendor's app or API.
  • LLM is a text model, Vision (VLM) also reads images, and Reasoning works through a problem step by step before it answers: slower and pricier per reply, but stronger on maths, code and planning. Image, video and audio generators carry their own badges.

Closed vendors do not publish parameter counts, which is why that cell is empty for them rather than a guess.

The leaders of each year

Above every year sits a strip with the models of that year that lead text, coding, web development, agents, image and video tasks today. This is a present-day view, not a reconstruction of the past: a model released in spring may have been overtaken by autumn, and the strip shows how its generation holds up now on the leaderboards.

Filters, editions and limits

You can narrow the view by vendor, country, open weights and type, or search for a name. Each year also has its own page, a yearly edition you can link to. Release dates come from model catalogues — models.dev, LiteLLM and Hugging Face — and mark the day a model appeared there, which may differ from the announcement by a few days. Models whose sources give only a year are not shown. The country is that of the company's headquarters, not of its data centres or its users.