Skip to content
Discuss your project
Model Change Index
  • OverviewNext shutdowns and latest changes
  • ShutdownsEvery date, one row per platform
  • ChangesReleases, deprecations, redirects and prices
  • LeaderboardEvery model, its scores and price
  • Check my stackYour models on one timeline
Track my stack

Model Change Index

Every model shutdown, deprecation, silent redirect and price change across 10 platforms, with the date on each and what breaks when you switch.

Sources checked · 445 sources

data.json Change feed

Next shutdown
TodayClaude Sonnet 4 · Amazon Bedrock
Shutdowns · next 30 days
2928 models on 5 platforms
Shutdowns · next 90 days
5343 models on 7 platforms
Silent redirects
9Model IDs answered by another model
Price changes · 90 days
819 releases in the last 30 days

Upcoming shutdowns

The soonest model shutdowns, one row per platform. Select a model for its dates on every platform.
DateInModelPlatformReplacement
TodayClaude Sonnet 4Amazon BedrockAmazon BedrockNone named
6 daysGemini 2.5 FlashGoogle Vertex AIGoogle Vertex AIGemini 3.8 Flash
6 daysGemini 2.5 Flash-LiteGoogle Vertex AIGoogle Vertex AIGemini 3.8 Flash
6 daysGemini 2.5 ProGoogle Vertex AIGoogle Vertex AIGemini 3.8 Flash
7 daysDeepSeek-R1Google Vertex AIGoogle Vertex AINone named
7 daysDeepSeek-V3.1Google Vertex AIGoogle Vertex AINone named
7 daysDeepSeek-V3.2Google Vertex AIGoogle Vertex AINone named
7 daysGLM-4.7Google Vertex AIGoogle Vertex AINone named
7 daysGLM-5Google Vertex AIGoogle Vertex AINone named
7 daysgpt-oss-20bGoogle Vertex AIGoogle Vertex AINone named
7 daysKimi K2 ThinkingGoogle Vertex AIGoogle Vertex AINone named
7 daysLlama 3.3 70B InstructGoogle Vertex AIGoogle Vertex AINone named
17 more in this window. All 140 firm shutdown dates

Latest changes

All 156
  1. ShutdownAmazon BedrockClaude Sonnet 4 reaches end of life on Amazon Bedrock
  2. ReleaseGemini API, Google Vertex AIGemini 4 Argon released
  3. ReleaseGoogle Vertex AIGemini 4 Argon Cyber Permissive released for allowlisted customers
  4. RedirectGemini APIGemini 3.5 Flash requests now go to Gemini 3.6 Flash
  5. RedirectGemini APIGemini 3.7 Flash requests now go to Gemini 3.8 Flash
  6. ReleaseClaude API, Google Vertex AIClaude Haiku 5.5 released
  7. ReleaseAmazon BedrockClaude Haiku 5.5 available on Amazon Bedrock
  8. API changeOpenAI APIchat-latest updated again
  9. ReleaseMistral APIMistral Large 4 in Public Preview on the Mistral API
  10. Price changeFireworks AIDeepSeek V4.1 Flash serverless prices change on Fireworks
  11. ShutdownMistral APILeanstral 1.5 retires from the Mistral API

Shutdown calendar · next 6 months

Firm dateEarliest possible dateDot size: models that day

Nov 2026Dec 2026Jan 2027Feb 2027Mar 2027Apr 2027TodayOpenAI APIOct 23, 2026, OpenAI API: GPT-3.5 Turbo, GPT-4, GPT-4 Turbo, GPT-4.1 nano, GPT-4o, o1 and 3 more9Dec 11, 2026, OpenAI API: GPT-5, GPT-5 Mini, GPT-5 nano, GPT-5 Pro, o3, o3-pro6Jan 14, 2027, OpenAI API: GPT-4.1, GPT-4.1 Mini, GPT-4o, GPT-4o, GPT-4o Mini5Apr 1, 2027, OpenAI API: GPT-5.1, GPT-5.3-Codex, GPT-5.4 nano3Claude APIOct 15, 2026, Claude API: Claude Haiku 4.5 (earliest possible date)Nov 24, 2026, Claude API: Claude Opus 4.5 (earliest possible date)Nov 30, 2026, Claude API: Claude Sonnet 4.5Feb 5, 2027, Claude API: Claude Opus 4.6 (earliest possible date)Feb 17, 2027, Claude API: Claude Sonnet 4.6 (earliest possible date)Azure AI FoundryNov 19, 2026, Azure AI Foundry: o1, o1-pro, o3, o3-deep-research, o3-mini, o3-pro and 1 more7Apr 14, 2027, Azure AI Foundry: GPT-4.1, GPT-4.1 Mini, GPT-4.1 nano, GPT-4o, GPT-4o, GPT-4o Mini6Nov 15, 2026, Azure AI Foundry: Claude Haiku 4.5, Claude Haiku 4.5, codex-mini-latest3Feb 9, 2027, Azure AI Foundry: GPT-5, GPT-5 Mini, GPT-5 nano3Apr 6, 2027, Azure AI Foundry: Grok 4.20 (Non-Reasoning), Grok 4.20 (Reasoning) (earliest possible date)Oct 16, 2026, Azure AI Foundry: Command A Plus (earliest possible date)Nov 24, 2026, Azure AI Foundry: Claude Opus 4.5Nov 30, 2026, Azure AI Foundry: Claude Sonnet 4.5Dec 2, 2026, Azure AI Foundry: Chat Latest (earliest possible date)Dec 3, 2026, Azure AI Foundry: DeepSeek-V4-Flash (earliest possible date)Dec 9, 2026, Azure AI Foundry: GPT-4oJan 26, 2027, Azure AI Foundry: Kimi K2.5 (earliest possible date)Feb 2, 2027, Azure AI Foundry: Claude Opus 4.6Feb 10, 2027, Azure AI Foundry: Claude Sonnet 4.6Mar 17, 2027, Azure AI Foundry: GPT-5-CodexApr 2, 2027, Azure AI Foundry: Claude Mythos Preview (earliest possible date)Apr 6, 2027, Azure AI Foundry: Claude Opus 4.7Apr 7, 2027, Azure AI Foundry: GPT-5 ProAmazon BedrockMar 30, 2027, Amazon Bedrock: DeepSeek-R1, DeepSeek-V3.1, Devstral 2, Gemma 3 12B IT, Gemma 3 27B IT, Gemma 3 4B IT and 20 more26Dec 2, 2026, Amazon Bedrock: Amazon Nova 2 Lite, Ministral 3 14B, Ministral 3 3B, Ministral 3 8B, Mistral Large 3 (earliest possible date)5Mar 31, 2027, Amazon Bedrock: Gemma 4 26B A4B, Gemma 4 31B, Gemma 4 E2B IT (earliest possible date)3Oct 29, 2026, Amazon Bedrock: gpt-oss-safeguard-120b, gpt-oss-safeguard-20b (earliest possible date)Nov 26, 2026, Amazon Bedrock: Jamba 1.5 Large, Jamba 1.5 MiniOct 14, 2026, Amazon Bedrock: Claude Sonnet 4Oct 16, 2026, Amazon Bedrock: Claude Haiku 4.5 (earliest possible date)Nov 24, 2026, Amazon Bedrock: Claude Opus 4.5 (earliest possible date)Dec 1, 2026, Amazon Bedrock: DeepSeek-V3.2 (earliest possible date)Dec 15, 2026, Amazon Bedrock: NVIDIA Nemotron 3 Nano 30B A3B (earliest possible date)Dec 16, 2026, Amazon Bedrock: Mistral Small 1.0 (earliest possible date)Jan 8, 2027, Amazon Bedrock: Claude Opus 4.1Jan 21, 2027, Amazon Bedrock: Palmyra X5 (earliest possible date)Jan 27, 2027, Amazon Bedrock: Kimi K2.5 (earliest possible date)Feb 4, 2027, Amazon Bedrock: Qwen3-Coder-Next (earliest possible date)Feb 5, 2027, Amazon Bedrock: Claude Opus 4.6 (earliest possible date)Feb 11, 2027, Amazon Bedrock: GLM-5 (earliest possible date)Feb 12, 2027, Amazon Bedrock: MiniMax-M2.5 (earliest possible date)Feb 17, 2027, Amazon Bedrock: Claude Sonnet 4.6 (earliest possible date)Mar 11, 2027, Amazon Bedrock: NVIDIA Nemotron 3 Super 120B A12B (earliest possible date)Mar 26, 2027, Amazon Bedrock: Palmyra Vision 7B (earliest possible date)Apr 8, 2027, Amazon Bedrock: Claude Sonnet 4.5Google Vertex AIOct 21, 2026, Google Vertex AI: DeepSeek-R1, DeepSeek-V3.1, DeepSeek-V3.2, GLM-4.7, GLM-5, gpt-oss-20b and 7 more13Oct 20, 2026, Google Vertex AI: Gemini 2.5 Flash, Gemini 2.5 Flash-Lite, Gemini 2.5 Pro3Oct 15, 2026, Google Vertex AI: Claude Haiku 4.5 (earliest possible date)Nov 19, 2026, Google Vertex AI: Gemini 3.6 FlashNov 24, 2026, Google Vertex AI: Claude Opus 4.5 (earliest possible date)Dec 24, 2026, Google Vertex AI: Claude Sonnet 5 (earliest possible date)Jan 24, 2027, Google Vertex AI: Claude Opus 5 (earliest possible date)Jan 28, 2027, Google Vertex AI: Gemini 3.7 FlashFeb 5, 2027, Google Vertex AI: Claude Opus 4.6 (earliest possible date)Mar 1, 2027, Google Vertex AI: Claude Fable 5.1 (earliest possible date)Mistral APIOct 31, 2026, Mistral API: GLM-5.2Together AIOct 22, 2026, Together AI: DeepSeek-V4-Flash, Llama 3.3 70B Instruct

What breaks: the changes a new model name hides

  • Parameters

    temperature, top_p and top_k are ignored. Google says later model generations will return an error.

    Gemini 3.6 Flash, Gemini 3.5 Flash-Lite · Gemini API ·

  • Parameters

    A request whose last non-empty turn is a model turn returns a 400 error.

    Gemini 3.6 Flash, Gemini 3.5 Flash-Lite · Gemini API ·

  • Parameters

    Custom temperature, top_p and top_k values are ignored.

    Gemini 3.6 Flash, Gemini 3.5 Flash-Lite · Gemini API, Google Vertex AI ·

  • Parameters

    On Agent Platform, custom frequency and presence penalty values return an API error.

    Gemini 3.6 Flash, Gemini 3.5 Flash-Lite · Gemini API, Google Vertex AI ·

  • Parameters

    A request whose last turn is a model turn returns an error.

    Gemini 3.6 Flash, Gemini 3.5 Flash-Lite · Gemini API, Google Vertex AI ·

  • Defaults

    Gemini 3.5 Flash-Lite defaults to the minimal thinking level; set medium or high for agent work.

    Gemini 3.6 Flash, Gemini 3.5 Flash-Lite · Gemini API, Google Vertex AI ·

  • Behavior

    Magistral models always returned thinking traces. Mistral Small 4 and Mistral Medium 3.5 reason through the reasoning_effort parameter: high returns a thinking chunk before the answer, none omits it.

    Devstral 2, Magistral Medium 1.2, Mistral Nemo 12B Instruct · Mistral API ·

  • Availability

    Inference requests to retired deployments return 410 Gone.

    Claude Opus 4.1 to Claude Opus 5 · Azure AI Foundry ·

Built by TeqEngine. Forward-deployed AI engineering, no model to sell. Moving an agent to a new model? Ask for a Next-Model Check.