AI API Deadlines: Model Shutdowns and Price Changes
Provider pages announce a shutdown once and then bury it. This is the list in date order, each row with its source and the first thing to do, so a retirement does not arrive as a failing background job. The next one is Fri, 23 Oct 2026.
October 2026
- ShutdownOpenAI
o1, o1 pro, o3-mini, o4-mini, GPT-4, GPT-4 Turbo, GPT-3.5 Turbo 0125, GPT-4.1 nano and gpt-image-1 are removed, with their fine tuned versions
Do this: Search your code and config for these model IDs today, switch each to a tested replacement, and rerun your evals before the date.
- ShutdownGoogle
The Gemini Deep Research agent deep-research-pro-preview-12-2025 shuts down
Do this: Change the agent parameter in your interactions.create calls to deep-research-preview-04-2026 or deep-research-max-preview-04-2026.
- Read onlyOpenAI
Existing evals in the OpenAI Evals platform become read only
Do this: Export your eval definitions and results now, because the dashboard and API shut down on 30 November.
November 2026
- ShutdownGitHub
GitHub Copilot removes Claude Opus 4.8, Claude Opus 5 and GPT-5.6 Terra from every Copilot experience
Do this: Move saved workflows and agent configs to Claude Opus 5.5, GPT-6.1 Sol or GPT-6 Luna. Annual individual plans keep Opus 4.8.
- PriceOpenAIAt least through
GPT-5.6 Sol promotional pricing of $4 input and $20 output per million tokens is guaranteed only through this date
Do this: Model your bill at the earlier price the page implies, $5 and $30, until OpenAI says the promotion is permanent.
- ShutdownAnthropic
Claude Sonnet 4.5 (claude-sonnet-4-5-20250929) is retired on the Claude API
Do this: Switch to claude-sonnet-5-5, which Anthropic names as the replacement, and check your token counts again because the tokenizer changed.
- ShutdownOpenAI
Agent Builder, the Evals dashboard and API, and the v1/prompts reusable prompts API shut down
Do this: Move prompt text into your own code and replace Agent Builder flows with the Agents SDK before the date.
December 2026
- PriceGoogle
Gemini 4 Argon moves from $2 input and $10 output to $4 and $20 per million tokens
Do this: Cost your product at the December price, not the launch price.
- ShutdownOpenAI
gpt-image-1-mini, gpt-image-1.5 and chatgpt-image-latest are removed
Do this: Move image calls to gpt-image-2.5-sunburst or gpt-image-2.5-flare, and compare output quality and cost on your own prompts.
- ShutdownOpenAI
Older GPT-5 and o3 snapshots are removed: gpt-5, gpt-5-mini, gpt-5-nano, gpt-5-pro, o3 and o3-pro
Do this: Pin the replacement OpenAI names for each snapshot and run your evals before the date.
January 2027
- ShutdownOpenAI
tts-1, tts-1-hd and both gpt-4o-mini-tts snapshots are removed, and active customers can no longer create fine tuning jobs
Do this: Move speech output to gpt-realtime-2.1-mini, and finish any fine tuning you still need to run.
- ShutdownOpenAI
GPT-4o, GPT-4o mini, GPT-4.1 and GPT-4.1 mini are removed, with their fine tuned versions
Do this: Test gpt-6-luna, the replacement OpenAI names for all four, against your current quality bar.
- ShutdownOpenAI
The older audio and realtime family is removed, including gpt-realtime, gpt-audio and their 4o and mini variants
Do this: Move to gpt-realtime-2.1, gpt-realtime-2.1-mini or gpt-audio-1.5, as OpenAI maps each one.
February 2027
- ShutdownOpenAI
whisper-1, gpt-4o-transcribe, gpt-4o-mini-transcribe and gpt-4o-transcribe-diarize are removed
Do this: Move transcription to gpt-transcribe or gpt-live-transcribe, and move translation to v1/realtime/translations.
April 2027
- ShutdownOpenAI
gpt-5.3-codex, gpt-5.1 and gpt-5.4-nano are removed
Do this: Plan the move to gpt-6-sol for the first two and gpt-6-luna for gpt-5.4-nano.
Questions
Where do the dates come from?
From each provider's own page: OpenAI's deprecations page and model pages, Anthropic's model deprecations page, Google's Gemini API release notes and pricing page, and GitHub's changelog. Every row links to its source. A date that only appears in a third party summary is not listed.
How do I get these in my calendar?
Subscribe to https://pulse.cafiyn.com/signals/deadlines.ics. In Google Calendar choose Other calendars, then From URL. In Apple Calendar choose File, then New Calendar Subscription. In Outlook choose Add calendar, then Subscribe from web. Each event carries a reminder one week before the date.
How much notice does OpenAI give before it retires a model?
OpenAI's deprecations page says generally available models get at least six months, specialised variants such as chat, Codex and deep research variants at least three months, and preview models may get as little as two weeks. It also says a faster timeline is possible for safety or compliance reasons.
Can I check shutdown dates from code?
For OpenAI, yes. The List models endpoint returns an optional shutdown_date field on each model, as a YYYY-MM-DD string or null when none has been announced. The same field is on the Retrieve a model endpoint.
How often is this updated?
With every weekly issue of Startup Signals, and whenever a provider announces a new date between issues. The page itself refreshes hourly, so rows move from upcoming to passed on their own.
What is the difference between a shutdown and a read only date?
A shutdown is the day a model or endpoint stops accepting requests. A read only date is the day existing data, such as saved evals, can still be viewed but no longer changed, ahead of a later shutdown of the product itself.
Open the tool.
Most migrations land on a model with a different price. The Cost Comparator shows what your own traffic costs on each one.
Price the replacement before you switch