Text & reasoning

Everything that shipped

• 3dTEXTClaude FableAnthropicUSNEW DROPClaude Fable 5.2 is listed on OpenRouter.
  5.2
• 6dTEXTLongCatMeituanCNNEW DROPLongCat 2.5 is listed on Hugging Face.
  2.5
• 2wTEXTGPT AstraOpenAIUSNEW DROPGPT-6.1 Terra brings fast and capable computer use to everyone.
  6.1T
• 2wTEXTKimiMoonshot AICNNEW DROPKimi K3.1 Pro and Highspeed launch across Kimi apps, Code, and API.
  K3.1P
• 3wTEXTDeepSeek V4.1DeepSeekGlobalNEW DROPDeepSeek V4.1-Flash-1017 lands with integrated Search & Fetch APIs and sub-second reasoning.
  4.1
• 3wTEXTMiniMax MMiniMaxCNNEW DROPMiniMax releases MiniMax-M3.1-Flash multimodal open weights on Hugging Face.
  3.1
• 3wTEXTClaude HaikuAnthropicUSNEW DROPClaude Haiku 5.5 ships with 1M context window and near-Opus coding speed.
  5.5
• 4wTEXTLing FlashAnt GroupCNNEW DROPLing 3.1 Flash expands reasoning performance and instruction adherence.
  3.1
• 4wTEXTGemini ArgonGoogle DeepMindUSNEW DROPGemini 4 Argon opens to developers, enterprises and consumers.
  4
• 1moTEXTMistral LargeMistral AIEUMistral Large 4: a 1-trillion-parameter natively multimodal MoE.
  4
• 1moTEXTGemini ArgonGoogle DeepMindUSGemini 4 Argon announced, rolling out first to trusted cyber defenders.
  4 Pre
• 1moTEXTClaude OpusAnthropicUSClaude Opus 5.5 takes #1 on frontier reasoning and autonomous software engineering.
  5.5
• 1moTEXTGPT AstraOpenAIUSGPT-6.1 Sol offers near-Astra intelligence at one-fifth of Astra’s API token prices.
  6.1S
• 1moTEXTClaude SonnetAnthropicUSClaude Sonnet 5.5 is listed on OpenRouter and models.dev.
  5.5
• 1moTEXTGPT AstraOpenAIUSGPT-6 Sol and Luna bring frontier intelligence to everyday work.
  6S
• 1moTEXTGrokxAIGlobalGrok 4.7 is listed on OpenRouter and models.dev.
  4.7
• 2moTEXTClaude FableAnthropicUSFable 5.1 expands multi-agent alignment and complex scientific theorem proving.
  5.1
• 2moTEXTQwenAlibabaCNQwen3.8 Max 0902 checkpoint introduces refined multi-turn tool-use.
  0902
• 2moTEXTGPT AstraOpenAIUSGPT-6 Astra debuts as OpenAI’s most intelligent and aligned model yet.
  6.0
• 2moTEXTGemini FlashGoogle DeepMindUSGemini 3.8 Flash and 3.8 Flash Cyber arrive three weeks after 3.7.
  3.8
• 2moTEXTGLMZ.aiCNZ.ai releases GLM-5.3 and GLM-5.3-Flash open-weight checkpoints on Hugging Face.
  5.3
• 2moTEXTDeepSeek R1DeepSeekCNDeepSeek R1 matches frontier reasoning with open weights and emergent self-verification chains.
  R1
• 2moTEXTGemini FlashGoogle DeepMindUSGemini 3.7 Flash, Google’s most intelligent workhorse model yet for coding and agents.
  3.7
• 3moTEXTGrokxAIGlobalGrok 4.6 is listed on OpenRouter and models.dev.
  4.6
• 3moTEXTQwenAlibabaCNQwen3.8-Max, 2.4T parameters and the first Max tier to open.
  3.8
• 3moTEXTDeepSeek V4.1DeepSeekGlobalDeepSeek V4 Flash delivers high-throughput reasoning at record-low inference cost.
  4
• 3moTEXTKimiMoonshot AICNFull K3 weights published on Hugging Face, on the date Moonshot committed to.
  K3
• 3moTEXTQwenAlibabaCNQwen3.7-Flash lands between the 3.5 train and 3.8-Max.
  3.7
• 3moTEXTClaude OpusAnthropicUSOpus 5 lands below Fable 5, at half the price.
  5
• 3moTEXTLing FlashAnt GroupCNLing-3.0-flash released with open weights.
  3.0
• 3moTEXTQwenAlibabaCNThree Qwen releases inside 72 hours.
  3.5
• 3moTEXTGPT-5.6OpenAIUSGPT-5.6 arrives as a lineup: Sol, Terra and Luna.
  5.6
• 3moTEXTGemini FlashGoogle DeepMindUSGemini 3.6 Flash ships as a three-model trio.
  3.6
• 3moTEXTLongCatMeituanCNLongCat 2.0, a 1.6T-parameter sparse MoE from Meituan.
  2.0
• 3moTEXTGrokxAIGlobalGrok 4.5 pushes coding and knowledge work at lower token use.
  4.5
• 3moTEXTKimiMoonshot AICNKimi K3 launches at 2.8T parameters with native vision and a 1M-token context.
  K3
• 4moTEXTHy3TencentGlobalHy3 ships with configurable reasoning.
  3
• 4moTEXTClaude SonnetAnthropicUSSonnet 5 arrives under the Opus 5 flagship.
  5
• 4moTEXTGLMZ.aiCNGLM-5.2 ships 753B MIT-licensed open weights on Hugging Face.
  5.2
• 5moTEXTClaude FableAnthropicUSFable 5 ships as Anthropics premier Mythos-class intelligence engine.
  5
• 5moTEXTMiniMax MMiniMaxCNMiniMax-M3 ships 427B multimodal MoE weights on Hugging Face.
  3.0
• 6moTEXTQwenAlibabaCNQwen 3 generation opens with hybrid reasoning.
  3.0
• 6moTEXTHy3TencentGlobalHy3 preview, ahead of the full release.
  3-pre
• 6moTEXTKimiMoonshot AICNKimi K2.6 open-sources long-horizon coding and agent swarms.
  K2.6
• 7moTEXTClaude OpusAnthropicUSIncremental bump to the 4-series flagship.
  4.8
• 7moTEXTMiniMax MMiniMaxCNMiniMax-M2.7 open-weight 228B MoE lands on Hugging Face.
  2.7
• 8moTEXTGemini FlashGoogle DeepMindUSGemini 3 generation reaches the Flash tier.
  3.0
• 8moTEXTGPT-5.6OpenAIUS5.2 tightens routing between fast and thinking modes.
  5.2
• 9moTEXTKimiMoonshot AICNKimi K2.5 turns K2 into a native multimodal model.
  K2.5
• 11moTEXTClaude OpusAnthropicUSOpus 4.5 lands with a large price cut.
  4.5
• 1yTEXTGPT-5.6OpenAIUSGPT-5 replaces the 4-series across ChatGPT.
  5.0
• 1yTEXTClaude OpusAnthropicUSOpus 4.1 refines agentic search and coding.
  4.1
• 1yTEXTKimiMoonshot AICNK2 established the open MoE line.
  K2
• 1yTEXTGrokxAIGlobalGrok 4 released with a heavy multi-agent tier.
  4.0
• 2yTEXTMistral LargeMistral AIEUMistral Large 2 delivers 123B parameter frontier reasoning with 128k context.
  2