OpenAI: "The model … has been deprecated" or "Model not found" after a shutdown
The model `computer-use-preview-2025-03-11` has been deprecated
Short answer: OpenAI has shut down the model your request names, so the API refuses it. Change the model ID to a current one (the deprecations page names the replacement for each), test, and redeploy. The gpt-4o, gpt-4o-mini, gpt-4.1 and gpt-4.1-mini snapshots and their fine-tunes shut down on 14 January 2027, gpt-5, gpt-5-mini and o3 on 11 December 2026.
Why it happens
OpenAI retires models on announced dates and, after the shutdown, requests that name them fail. Its error-code guide does not document a shutdown-specific message. On 1 August 2026, the day after a shutdown, developers saw Chat Completions answer "The model … has been deprecated" and the Responses API answer "Model not found …"; an ID OpenAI does not recognise returns 404 model_not_found. The 14 January 2027 notice lists dated snapshots (for example gpt-4o-mini-2024-07-18); the aliases gpt-4o-mini and the others point only to those snapshots.
You may also see
Model not found computer-use-preview-2025-03-11404 model_not_found: The model <id> does not exist or you do not have access to it.
Fix
- Find the model ID in the failing request: your code, a
.envfile, an n8n node, a plugin setting or a Make module. The AI model retirement scanner finds every retiring ID in pasted code, config or n8n workflow JSON. - Look up the replacement on OpenAI's deprecations page (gpt-4o, gpt-4o-mini, gpt-4.1, gpt-4.1-mini →
gpt-6-luna; gpt-5 and o3 →gpt-5.6-sol; gpt-5-mini →gpt-5.6-terra; whisper-1 →gpt-transcribe). - If the replacement is a reasoning model such as gpt-6-luna: on Chat Completions with tools set
reasoning_efforttononeor use the Responses API, and removetemperatureandtop_punless effort isnone. - For a fine-tuned model, retrain on a current base: fine-tunes stop with their base snapshot, and new fine-tuning jobs end on 6 January 2027.
- Test with real inputs, then redeploy every service, workflow and plugin that uses the ID.
Watch out
- A 404 model_not_found can also mean the ID is misspelled or your project has no access to it; check the ID against OpenAI's model list before blaming a shutdown.
- Hosted tools change at their own pace: n8n's OpenAI node hardcodes whisper-1 for transcription, and Make moved some scenarios onto gpt-4o-2024-08-06 and gpt-4.1, which also retire on 14 January 2027.
Sources
- OpenAI: Deprecations (shutdown dates, snapshots, replacements)
- OpenAI community: errors seen on 1 Aug 2026, the day after a shutdown
- OpenAI: Error codes
- OpenAI: gpt-6-luna (reasoning effort, function calling)