Four GPT-4 series models and their fine-tunes stop working in the API on 14 January 2027, with gpt-6-luna named as the replacement for all of them; text-to-speech and Whisper retirements follow close behind.
AI API retirements: Claude Sonnet 4.5 retires on 30 November 2026; three OpenAI models leave the API on 1 April 2027
EditorialBy TrustList Editorial
Anthropic retires Claude Sonnet 4.5 from its API on 30 November 2026. OpenAI will remove GPT-5.3-Codex, GPT-5.1 and GPT-5.4-Nano on 1 April 2027, and some text-to-speech models on 6 January 2027. Each names a replacement.
About AI API retirements: Claude Sonnet 4.5 retires on 30 November 2026; three OpenAI models leave the API on 1 April 2027
AI API retirements: Claude Sonnet 4.5 retires on 30 November 2026; three OpenAI models leave the API on 1 April 2027
1 October 2026: Two AI model providers have set retirement dates for models that many business applications still call. Anthropic notified developers on 30 September 2026 that Claude Sonnet 4.5 will be retired from its API on 30 November 2026. OpenAI announced on 1 October 2026 that three models will be removed from its API on 1 April 2027, and that some text-to-speech models will be removed on 6 January 2027.
Not yet independently verified. Both sources are the providers' own documentation; no independent report of these specific notices had been found at 2026-10-05. The full list of affected text-to-speech models was not read. We will update this when it can be confirmed, and remove this note.
The dates
Anthropic. According to its deprecation page, claude-sonnet-4-5-20250929 was deprecated on 30 September 2026 and retires on 30 November 2026, with claude-sonnet-5-5 as the recommended replacement. That is two months' notice.
OpenAI. Its deprecations page lists, with six months' notice:
gpt-5.3-codex: removed 1 April 2027, replacementgpt-6-solgpt-5.1: removed 1 April 2027, replacementgpt-6-solgpt-5.4-nano: removed 1 April 2027, replacementgpt-6-luna
A second notice of the same day says certain text-to-speech models will be removed on 6 January 2027, with migration to gpt-realtime-2.1-mini.
Who is affected
Any company whose products, internal tools or automations call these model identifiers directly, including customer-support bots, coding tools, document processing and voice applications. Software vendors who embed these models in their own products must migrate before their customers notice.
What to do
- Search code, configuration and prompt-management tools for the exact model identifiers above.
- Test the replacement models on your own tasks before switching: output quality, format, latency and cost per task can all change.
- Check pricing for the replacements and update cost forecasts.
- Put the dates in your release plan: 30 November leaves little time for applications on Claude Sonnet 4.5.
- Ask your SaaS vendors which models their AI features use and how they will handle these retirements.
Why it matters for buyers
Model retirements are now a routine part of buying AI. Notice periods vary from weeks to six months between providers. When choosing an AI vendor or building on one, check the published deprecation policy and how much notice it guarantees, and keep model choice configurable rather than hard-coded.
Sources
- Anthropic (Claude Developer Platform): Model deprecations - 30 September 2026
- OpenAI API documentation: Deprecations - 1 October 2026
Categories & features
TrustList Weekly
The week in software and IT, in one email
The news that matters to buyers, new rankings and our own research. Every Thursday, free, and easy to leave.
You may be interested in
OpenAI planned to halve ChatGPT Pro 200’s included usage from 30 October 2026, then cancelled the cut on 29 October and raised lower-allowance Pro 200 subscriptions to the higher level. The $500-a-month Pro 500 plan with Ultrafast remains.
From 31 October new sign-ups lose access to kimi-k2.6 and kimi-k2.7-code; existing API users keep them until a retirement Moonshot plans for the end of December, and K3 bills cache writes separately.
Agent tool calls through DeepSeek Harness, Claude Code or Codex already use the new Search and Fetch endpoints for free; per-request billing starts at midnight Beijing time on 1 November.
The San Francisco startup, from Y Combinator's Spring 2026 batch, builds patient-level datasets from scattered hospital records and works in the US, Brazil, India and Argentina.
The registration statement sets no price or share count yet. It shows revenue of $16.3 billion in the first half of 2026 and a governance structure that keeps a trust in charge of some board seats.
Snorkel AI announced a $350 million Series E on 22 September 2026, co-led by Insight Partners and S32, at a valuation it puts at $3.5 billion. The money goes into its "agentic data factory", which supplies training data and environments to AI labs.
Cloudflare says it will begin billing for Artifacts, its Git-compatible versioned storage for AI agents in open beta on the Workers Paid plan, on 12 November 2026, not the 15 October first announced. Pricing is per repository operation and per GB stored.
More on TrustList
Everything here links back to the same verified catalogue. Pick your next stop.
- More United StatesThe ranking for this subject
- CompaniesAgencies, consultancies and IT service providers, ranked by verified reviews.
- ProductsSoftware and SaaS with pricing, features, integrations and alternatives.
- AwardsAnnual recognition decided by verified reviews and an independent jury.
- LaunchesNew products and releases, voted up by the community every day.
- AI ModelsBenchmark scores and community ratings for every major model.
- RequestsBuyers describe what they need; vendors respond directly.
- PeopleReviewers, authors and makers with public profiles.
- ComparePut up to four listings side by side before you shortlist.