LLM Deprecation Calendar
Every announced shutdown date across OpenAI · Anthropic · Google · AWS Bedrock · Azure Foundry · Mistral - chronological. Last verified: 2026-07-20. Updated weekly.
Corrections made 2026-07-20: the OpenAI Jul 23 / Oct 23 waves were listed the wrong way round (Jul 23 is the current-generation Codex and chat-latest tier, not the gpt-3.5/gpt-4/o1 tier), and mistral-document-ai-2505 was listed as Aug 17 when Azure retires it Jul 20. Both fixed against the sources today. See the Week 30 digest for what changed and why it matters.
How to read this: every date comes from the vendor's official page (linked per row). Google marks its dates as "earliest possible" - treat them as floors. OpenAI groups some shutdowns into waves (Jul 23 / Oct 23 and Mar 26 / Sep 28) and the wave order does not follow model age - verify your exact model ID on their deprecations page rather than assuming the older thing dies first. This calendar is maintenance, not legal advice: always confirm at the source before migrating.
July 2026
| Date | Provider | What retires | Migrate to | Src |
|---|
| Jul 20 | Azure Foundry | mistral-document-ai-2505 (preview) | mistral-document-ai-2512 / mistral-ocr-4-0 | ✓ |
| Jul 21 | Anthropic | claude-mythos-preview (Claude API; the Azure Foundry listing runs to 2027-04-02) | claude-mythos-5 | ✓ |
| Jul 22 | Anthropic | old memory-listing semantics under managed-agents-2026-04-01 header | adapt list/pagination code (see W29) | ✓ |
| Jul 23 | OpenAI | shutdown wave 1 of the 2026-04-22 announcement - 18 current-generation IDs, not the old ones: gpt-5-chat-latest, gpt-5-codex, gpt-5.1-chat-latest, gpt-5.1-codex, gpt-5.1-codex-max, gpt-5.1-codex-mini, gpt-5.2-codex, computer-use-preview (+ dated), o3-deep-research, o4-mini-deep-research (+ dated), gpt-4o-search-preview-2025-03-11, gpt-4o-mini-search-preview-2025-03-11, gpt-4o-mini-tts-2025-03-20, gpt-audio-mini-2025-10-06, gpt-realtime-mini-2025-10-06. The gpt-3.5-turbo/gpt-4/o1/o3-mini group is wave 2 (Oct 23) | gpt-5.5 · gpt-5.4-mini · gpt-5.5-pro | ✓ |
| Jul 23 | Azure Foundry | gpt-realtime-mini (2025-10-06) | version 2025-12-15 | ✓ |
| Jul 24 | Anthropic | fast mode (speed:"fast") on Opus 4.7 - errors after this date | fast mode on Opus 4.8 | ✓ |
| Jul 30 | AWS Bedrock | EOL: Claude 3 Sonnet; Claude 3.5 Sonnet (20240620, select regions + GovCloud); 3.5 Sonnet v2 (AP regions); 3.7 Sonnet (GovCloud) | Sonnet 4.6 / Sonnet 5 | ✓ |
| Jul 30 | AWS Bedrock | Agents Classic closed to new customers; model catalog frozen | AgentCore | ✓ |
| Jul 31 | Mistral | Mistral Small 3.2 · Devstral 2 · Magistral Medium/Small 1.2 · Mistral Nemo 12B | Small 4 / Medium 3.5 / Ministral 3 8B | ✓ |
| Jul 31 | Azure Foundry | Stable-Diffusion-3.5-Large, Stable-Image-Core, Stable-Image-Ultra | - | ✓ |
August 2026
| Date | Provider | What retires | Migrate to | Src |
|---|
| Aug 5 | Anthropic | claude-opus-4-1-20250805 (Claude API) | claude-opus-4-8 | ✓ |
| Aug 5 | Azure Foundry | claude-opus-4-1 (preview) | claude-opus-4-8 | ✓ |
| Aug 5 / 28 | Azure Foundry | gpt-chat-latest preview snapshots (05-05 → Aug 5; 05-28 & 06-24 → Aug 28) | newer snapshot | ✓ |
| Aug 10 | OpenAI | gpt-5.2-chat-latest, gpt-5.3-chat-latest | gpt-5.5 | ✓ |
| Aug 13 | Azure Foundry | DeepSeek-R1 (Legacy) | DeepSeek-V4-Pro | ✓ |
| Aug 15 | Azure Foundry | MAI-Image-2e, MAI-Image-2 (preview) | MAI-Image-2.5(-Flash) | ✓ |
| Aug 17 | Anthropic | legacy Workbench (platform.claude.com/workbench) + experimental prompt tools APIs: /v1/experimental/generate_prompt, /v1/experimental/improve_prompt, /v1/experimental/templatize_prompt - these endpoints return an error after removal | updated Workbench (no saved prompts/variables/evals); move prompt generation into your own code | ✓ |
| Aug 17 | Google Gemini API | Imagen 4 family (imagen-4.0-*) + Gemini 3 image models (earliest possible date) | Nano Banana 2 / gemini-3.1-flash-lite-image | ✓ |
| Aug 19 | AWS Bedrock | EOL: Cohere Command R, Command R+ | Cohere Command A | ✓ |
| Aug 26 | OpenAI | Assistants API - full shutdown | Responses API + Conversations API | ✓ |
| Aug 26 | AWS Bedrock | Jamba 1.5 Large/Mini enter extended access (price may rise; EOL Nov 26) | newer Jamba | ✓ |
| Aug 30 | Azure Foundry | model-router previews (2025-05-19, 2025-08-07) | model-router GA | ✓ |
| Aug 31 | Anthropic | price: Sonnet 5 intro pricing ends ($2/$10 → $3/$15 from Sep 1) | budget +50% inputs | ✓ |
| Aug 31 | Mistral | Mistral Medium 3.1, Medium 3 | Medium 3.5 | ✓ |
| Aug 31 | Azure Foundry | TimeGEN-1 (Nixtla); tsuzumi-7b (NTT) | TimeGPT; tsuzumi2 | ✓ |
September 2026
| Date | Provider | What retires | Migrate to | Src |
|---|
| Sep 10 | AWS Bedrock | EOL: Claude 3 Haiku (incl. GovCloud) | Haiku 4.5 | ✓ |
| Sep 14 | AWS Bedrock | EOL: Amazon Nova Premier, Nova Sonic | newer Nova | ✓ |
| Sep 15 | Azure Foundry | sora-2 (2025-12-08 preview) | - (see Sep 24) | ✓ |
| Sep 16 / 18 | Azure Foundry | o1 (2024-12-17); o1-pro (2025-03-19) | o3 / GPT-5.x | ✓ |
| Sep 21 | Azure Foundry | gpt-audio-mini (2025-10-06) | version 2025-12-15 | ✓ |
| Sep 24 | OpenAI | Videos API + sora-2/sora-2-pro - shutdown, no 1P successor | consider Veo 3.1 / Gemini Omni Flash | ✓ |
| Sep 28 | OpenAI | shutdown wave 2 of 2025-09-26 announcement (legacy GPT snapshots) | current GPT-5.x | ✓ |
| Sep 30 | Mistral | Leanstral 1.5 (labs-leanstral-1-5; short-cycle labs model) | - | ✓ |
| Sep 30 | AWS Bedrock | EOL: Nova Canvas, Nova Reel (v1:0, v1:1) | - | ✓ |
October 2026
| Date | Provider | What retires | Migrate to | Src |
|---|
| Oct 1 | Azure Foundry | gpt-4o (all 2024 versions); gpt-4o-mini; o3-mini | gpt-5.1 / gpt-4.1-mini / o4-mini | ✓ |
| Oct 8 | AWS Bedrock | Claude Opus 4.1: extended access starts (price may rise; EOL Jan 8, 2027) | Opus 4.8 | ✓ |
| Oct 13 | Azure Foundry | Cohere-command-a-plus-05-2026 (preview) | - | ✓ |
| Oct 14 | Azure Foundry | gpt-4.1, gpt-4.1-mini, gpt-4.1-nano | GPT-5.x | ✓ |
| Oct 14 | AWS Bedrock | EOL: Claude Sonnet 4 (all regions; extended access since Jul 14) | Sonnet 4.6 / 5 | ✓ |
| Oct 15 | Azure Foundry | gpt-4o-mini-transcribe/tts, gpt-4o-transcribe(-diarize) | versions 2025-12-15 | ✓ |
| Oct 16 | Google Gemini API | gemini-2.5-pro, gemini-2.5-flash (earliest possible date) | gemini-3.1-pro-preview / gemini-3.5-flash | ✓ |
| Oct 16 | Azure Foundry | o3, o4-mini (2025-04-16) | GPT-5.x | ✓ |
| Oct 19 | Azure Foundry | claude-haiku-4-5, claude-opus-4-5, claude-sonnet-4-5 (previews) | newer Claude on Foundry | ✓ |
| Oct 23 | OpenAI | shutdown wave 2 of the 2026-04-22 announcement - the genuinely old tier (24 IDs): gpt-3.5-turbo (+ -0125, completions), gpt-4/gpt-4-0613/gpt-4-turbo variants + completions, gpt-4-1106-preview, gpt-4.1-nano, gpt-4o-2024-05-13, o1, o1-pro, o3-mini, o4-mini (+ fine-tuned ft-o4-mini), gpt-image-1 | gpt-5.5 · gpt-5.4-mini/nano · gpt-5.5-pro · gpt-image-2 | ✓ |
| Oct 23 | Azure Foundry | gpt-image-1 (preview) | gpt-image-2 | ✓ |
| Oct 31 | OpenAI | Evals platform → read-only (full shutdown Nov 30) | - | ✓ |
November-December 2026
| Date | Provider | What retires | Migrate to | Src |
|---|
| Nov 15 | Azure Foundry | codex-mini (2025-05-16) | - | ✓ |
| Nov 26 | AWS Bedrock | EOL: Jamba 1.5 Large/Mini | - | ✓ |
| Nov 30 | OpenAI | shutdown: v1/prompts (Reusable Prompts), Evals platform, Agent Builder | prompts in app code; Agents SDK | ✓ |
| Nov 30 | AWS Bedrock | EOL: TwelveLabs Marengo Embed v2.7 | - | ✓ |
| Dec 1 | OpenAI | gpt-image-1-mini, gpt-image-1.5, chatgpt-image-latest | gpt-image-2 | ✓ |
| Dec 5-26 | Azure Foundry | gpt-5.1-codex-max (Dec 5); o3-pro (Dec 10); gpt-5.2 (Dec 12); tts/tts-hd/whisper + several -2025-12-15 versions (Dec 15); gpt-image-1.5 (Dec 16); o3-deep-research (Dec 26) | newer versions | ✓ |
| Dec 11 | OpenAI | snapshots: gpt-5 / gpt-5-mini / gpt-5-nano (2025-08-07), gpt-5-pro (2025-10-06), o3 (2025-04-16), o3-pro (2025-06-10) | gpt-5.5, gpt-5.4-mini/nano, gpt-5.5-pro | ✓ |
2027 (announced so far)
| Date | Provider | What retires | Migrate to | Src |
|---|
| Jan 6 | OpenAI | fine-tuning: no new jobs accepted (final stage of 2026-05-07 wind-down) | prompt engineering / distillation / other providers | ✓ |
| Jan 8 | AWS Bedrock | EOL: Claude Opus 4.1 | Opus 4.8 | ✓ |
| Feb 6 | Azure Foundry | gpt-5, gpt-5-mini, gpt-5-nano (2025-08-07) | GPT-5.x | ✓ |
| Feb 23 - Mar 18 | Azure Foundry | gpt-realtime-1.5 & gpt-audio-1.5; gpt-5.3-codex; gpt-audio/gpt-realtime; gpt-5.4(-pro); gpt-5-codex; gpt-5.4-mini/nano | newer versions | ✓ |
| Apr 7 - 23 | Azure Foundry | gpt-5-pro, gpt-image-1-mini; text-embedding-3-large/small, ada-002; gpt-image-2; gpt-5.5 | newer versions | ✓ |
| May 6 / 15 | Azure Foundry | gpt-realtime-2; gpt-5.1 + gpt-5.1-codex(-mini) | newer versions | ✓ |
| May 7 | Google Gemini API | gemini-3.1-flash-lite-preview | gemini-3.1-flash-lite | ✓ |
| Jun 25 / Jul 9 | Azure Foundry | gpt-realtime-2.1(+mini); gpt-5.6-sol/terra/luna | newer versions | ✓ |
| Oct 1-16 | Azure Foundry | fine-tuned deployments: gpt-4o(-mini); gpt-4.1 family; o4-mini | newer FT targets | ✓ |
Anthropic "not sooner than" floors
Anthropic commits models won't retire before these dates (these are floors, not scheduled shutdowns):
| Model | Retirement not before |
|---|
claude-sonnet-4-5-20250929 | 2026-09-29 |
claude-haiku-4-5-20251001 | 2026-10-15 |
claude-opus-4-5-20251101 | 2026-11-24 |
claude-opus-4-6 | 2027-02-05 |
claude-sonnet-4-6 | 2027-02-17 |
claude-opus-4-7 | 2027-04-16 |
claude-opus-4-8 | 2027-05-28 |
claude-fable-5 | 2027-06-09 |
claude-sonnet-5 | 2027-06-30 |
Also active without a shutdown date: non-default temperature/top_p/top_k return 400 on Opus 4.7+, Opus 4.8 and Sonnet 5. Known gap: Google Vertex AI publishes no model-retirement table for partner models - Vertex-specific dates are unverified here; Gemini API dates above come from ai.google.dev.