The AI models and agents shutting down in November 2026, and what replaces them

What shuts down in November, by date

Some of November’s shutdowns are models leaving one app while they stay everywhere else, some are features, and some are whole products. Here they are in order, as each company’s own page gives them on November 2:

Date	What goes	Where	What replaces it
Nov 2	Claude Sonnet 4.6, Claude Opus 4.6, GPT-OSS-120b	Google Antigravity	Gemini 3.8 Flash on free plans; Claude Sonnet 5.5 and Opus 5.5 on a paid Google AI Pro, Ultra or Enterprise plan
Nov 11	Jules turns read-only, then shuts down December 31	Google Jules	Export your data; Google points to Antigravity
Nov 12, proposed	OpenAI’s models	Cursor	Your own OpenAI API key, the Codex extension, or a gateway
Nov 15	Claude Haiku 4.5, codex-mini	Microsoft Foundry	None named
Nov 16	Claude Opus 4.8, Claude Opus 5, GPT-5.6 Terra	GitHub Copilot	Claude Opus 5.5; GPT-6.1 Sol or GPT-6 Luna
Nov 17	Five Gemini Live and text-to-speech preview models	Gemini API	gemini-3.8-live; gemini-3.8-flash-tts or gemini-3.8-flash-lite-tts
Nov 19	o1, o1-pro, o3, o3-pro, o3-mini, o3-deep-research, o4-mini	Azure OpenAI, in Microsoft Foundry	gpt-5.6-sol, or gpt-5.6-terra for o3-mini and o4-mini
Nov 24	Claude Opus 4.5	Microsoft Foundry	None named
Nov 30	Claude Sonnet 4.5	Claude API and Microsoft Foundry	claude-sonnet-5-5
Nov 30	Fast mode on Claude Opus 4.8	Claude Code and the Claude API	Fast mode on Opus 5.5
Nov 30	Agent Builder, Evals, reusable prompts	OpenAI API	Agents SDK or ChatGPT Workspace Agents; Promptfoo; your own code

Claude Code’s changelog announced the change to fast mode in version 2.1.316, on October 31. The Antigravity removal is today: on the morning of November 2, Google’s models page still listed the three models, marked “Will be removed on November 2, 2026.”

Claude: Opus 4.8’s fast mode, and Sonnet 4.5

Fast mode on Opus 4.8 ends November 30. Claude Code 2.1.316 “Deprecated fast mode for Opus 4.8 as of October 30, 2026; it will be removed on November 30, 2026. After removal, fast mode requests on Opus 4.8 return an error and do not fall back to standard Opus 4.8. Migrate to Opus 5.5 to keep the speedup.” The Claude API’s fast mode page carries the same warning for speed: "fast" requests to claude-opus-4-8. Fast mode itself stays on Opus 5.5, where it’s out of research preview and cheaper: $8 in and $40 out per million tokens, against $10 and $50 on Opus 4.8, according to Claude Code’s fast mode docs. On Pro, Max, Ultra, Team and Enterprise, fast mode is paid from usage credits either way, not from the plan’s limits. Opus 4.8 itself stays on the Claude API: Anthropic’s deprecations page lists it as active, with retirement “Not sooner than May 28, 2027.”

Sonnet 4.5 leaves the Claude API November 30. Anthropic notified its users on September 30 and names claude-sonnet-5-5 as the replacement. “Requests to models past the retirement date will fail.” The date covers the Claude API, Claude Platform on AWS, Claude Models on Google Cloud (Anthropic managed) and Microsoft Foundry, whose own schedule agrees. Amazon Bedrock and Google’s own managed Claude set their own dates.

The people it catches are the ones who never picked Sonnet 4.5. Claude Code’s model configuration says that on Microsoft Foundry, Claude Code’s default model “defaults to Sonnet 4.5,” and that its sonnet alias means Sonnet 4.5 there and on Bedrock and Google Cloud’s Agent Platform. To move, set ANTHROPIC_DEFAULT_SONNET_MODEL to a newer Sonnet your deployment serves, or pick it by its full name.

Opus 4.5 and Haiku 4.5 aren’t on Anthropic’s list. The Claude API shows both as active, with retirement “Not sooner than” November 24 and October 15, 2026. Anthropic promises “at least 60 days’ notice before model retirement,” by email and in its documentation, and the page shows no notice for either, so neither can leave the Claude API before January. Microsoft’s schedule disagrees for Foundry, where it lists Haiku 4.5 retiring November 15 and Opus 4.5 November 24, with no replacement named. If you run either there, plan for Microsoft’s dates.

GitHub Copilot: three models on November 16

GitHub’s notice of October 15 says it will deprecate Claude Opus 4.8, Claude Opus 5 and GPT-5.6 Terra “across all GitHub Copilot experiences (including Copilot Chat, inline edits, ask and agent modes, and code completions) on November 16th, 2026.” It suggests Claude Opus 5.5 for both Claude models and “GPT-6.1 Sol or GPT-6 Luna” for Terra.

One group keeps a model: “Claude Opus 4.8 will remain available to individual GitHub Copilot subscribers on annual plans so they retain an Opus offering.” On Copilot Business and Enterprise, the suggested replacements are switched on automatically “unless an administrator has turned off the global default or explicitly disabled the model”; if yours has, they can enable them in Copilot’s model policies. Nobody needs to remove the old models: “No action is required to remove the models after they have been deprecated.”

Antigravity, Jules and Cursor

Three apps lose something this month, and each has its own memo:

• Antigravity, November 2. Claude Sonnet 4.6, Opus 4.6 and GPT-OSS-120b go, leaving free and Google AI Plus accounts with Gemini 3.8 Flash. Claude Sonnet 5.5 and Opus 5.5 stay on a paid Google AI Pro subscription, not its trial, and on Ultra and Enterprise. Who keeps Claude.
• Jules, November 11. Google’s Jules Turndown says “You have until November 11, 2026 to continue finishing up sessions,” after which Jules is read-only until it winds down on December 31. You can export your settings, scheduled tasks and every session now. What to do before November 11.
• Cursor, November 12, proposed. OpenAI’s help article still reads: “We have proposed a transition period of November 12, 2026,” and “Cursor may choose to end access sooner.” Your own OpenAI key works only in Cursor’s local Chat and Agent. How to keep GPT there.

The APIs: OpenAI, Gemini and Azure

OpenAI, November 30. Three products on OpenAI’s deprecations page close that day. Agent Builder “is scheduled to shut down,” with a migration guide “to continue with the Agents SDK or ChatGPT Workspace Agents”; “ChatKit remains available.” The Evals dashboard and API shut down too; existing evals have been read-only since October 31, and OpenAI’s suggested path is Promptfoo. The v1/prompts API and reusable prompt objects go the same day, and OpenAI’s advice is to “move reusable prompt content into your application code.”

Gemini, November 17. Google’s deprecations page lists five preview voice models for that date: gemini-3.1-flash-tts-preview, gemini-3.1-flash-live-preview, gemini-2.5-flash-native-audio-preview-12-2025, gemini-2.5-flash-preview-tts and gemini-2.5-pro-preview-tts. Google says its dates “indicate the earliest possible dates on which a model might be retired,” and that once a model is shut down, “the endpoint is no longer available.”

Azure OpenAI, November 19. Microsoft retires o1, o1-pro, o3, o3-pro, o3-mini, o3-deep-research and o4-mini in Foundry, 22 days before OpenAI’s own API retires o3 and o3-pro. It names gpt-5.6-sol as the replacement, or gpt-5.6-terra for o3-mini and o4-mini. Azure’s codex-mini goes on November 15, with no replacement listed.

What follows in December

• December 1: gpt-image-1-mini, gpt-image-1.5 and chatgpt-image-latest leave OpenAI’s API, for gpt-image-2.5-sunburst or gpt-image-2.5-flare.
• December 11: OpenAI’s API removes gpt-5, gpt-5-mini, gpt-5-nano, gpt-5-pro, o3 and o3-pro, with their dated snapshots. The replacements are GPT-5.6 Sol, Terra and Luna, with Sol in pro mode for the two pro models.
• December 31: Jules shuts down.
• Some day in December: ChatGPT and Codex start sharing one usage allowance. Nothing shuts down, but OpenAI hasn’t given the day (our memo).

How to check what you use

1. In Copilot, open the model picker. If it’s set to Claude Opus 4.8, Opus 5 or GPT-5.6 Terra, switch now rather than on the day. On Business and Enterprise, ask your admin whether the replacements are enabled.
2. In Claude Code, a ↯ beside the prompt means fast mode is on. Run /model to see which model it’s on; if that’s Opus 4.8, switch to Opus 5.5 before November 30. On Foundry, Bedrock or Agent Platform, set ANTHROPIC_DEFAULT_SONNET_MODEL so sonnet and the default don’t point at Sonnet 4.5.
3. In your own code and settings, search for each model name in the table, including in environment variables and saved settings. In Anthropic’s Console, the Usage page’s Export gives you a CSV with usage “broken down by API key and model.” OpenAI’s GET /v1/models returns each model’s shutdown_date, or null if none has been announced.
4. If the model you’re losing was your reason to use an app, you may not need the app to keep it. Claude Opus 4.8 stays on the Claude API and, at standard speed, in Claude Code on a Claude plan. Enjoy runs Claude Code beside Codex and the other labs’ agents, each on the plan you already have, and lets you switch agents in the middle of a conversation; it doesn’t add a lab’s models to another lab’s account.

We’ll update this memo if any of these dates move.