OpenAI notified developers on October 14, 2026 that several GPT-4 series models will be removed from its API on January 14, 2027. The recommended replacement for all of them is gpt-6-luna.
Which models are affected
According to OpenAI's deprecations page, these snapshots shut down on January 14, 2027:
- GPT-4o:
gpt-4o-2024-08-06andgpt-4o-2024-11-20 - GPT-4o mini:
gpt-4o-mini-2024-07-18 - GPT-4.1:
gpt-4.1-2025-04-14 - GPT-4.1 mini:
gpt-4.1-mini-2025-04-14
OpenAI says fine-tuned versions of these models follow the same shutdown schedule as their base models.
An earlier deadline for some snapshots
A separate, earlier entry on OpenAI's deprecations page sets an October 23, 2026 shutdown for the older gpt-4o-2024-05-13 snapshot and for gpt-4.1-nano, with different recommended replacements. Developers should check exactly which model IDs their apps call.
ChatGPT users are not affected
This notice is about the API. GPT-4o, GPT-4.1 and GPT-4.1 mini were already removed from ChatGPT earlier this year. The change affects apps, internal tools, chatbots and automations that send requests to OpenAI's API with these model names.
Moving to gpt-6-luna
OpenAI recommends gpt-6-luna as the replacement for all four models. A few points to check before switching, based on OpenAI's documentation as summarized by migration guides:
- gpt-6-luna is a reasoning model with a default reasoning effort of medium. Setting
reasoning_efforttononeis the closest match to the older non-reasoning models. - Function calling on the Chat Completions API requires
reasoning_effort: "none"; otherwise, use the Responses API. - Parameters such as
temperatureandtop_pshould be removed when reasoning effort is notnone. - gpt-6-luna cannot be fine-tuned, so teams relying on fine-tuned GPT-4o or GPT-4.1 models will need a different approach.
Azure runs its own schedule
Microsoft's Azure OpenAI service has a separate retirement timeline. Its model retirement page lists April 14, 2027 for the same GPT-4o, GPT-4o mini, GPT-4.1 and GPT-4.1 mini versions.
What to do now
Search your code, configuration files and gateways for both dated snapshot names and plain aliases like gpt-4o and gpt-4.1, then test your workloads on the replacement well before January 14, 2027.
