• GPT-4o and GPT-4.1 shut down

GPT-4o and GPT-4.1 shut down on January 14, 2027

OpenAI's deprecations page gives January 14, 2027 as the shutdown date for gpt-4o-2024-08-06, gpt-4o-2024-11-20, gpt-4o-mini-2024-07-18, gpt-4.1-2025-04-14 and gpt-4.1-mini-2025-04-14. The notice went out on October 14, 2026, and the replacement it recommends for all five is gpt-6-luna. Two older snapshots, gpt-4o-2024-05-13 and gpt-4.1-nano, already shut down on October 23.

What breaks

Changing the model string is the easy half. gpt-6-luna is a reasoning model, and on Chat Completions it accepts function tools only when reasoning_effort is "none". The SDKs leave the default (medium), so a tool call that worked on GPT-4o fails after a plain swap. temperature and top_p are rejected for the same reason unless effort is "none".

The scanner flags the short names too (gpt-4o, gpt-4o-mini, gpt-4.1, gpt-4.1-mini) and fine-tunes built on these models. I ran the command-line version on 21 public repos and read every finding by hand before trusting it here.

How to fix it

Swap the ID, then go through each call that sends tools or sampling parameters: set reasoning_effort: "none" there, or move the call to the Responses API if you want the model to reason. Run a sample of your real prompts on gpt-6-luna before the date, because answers and token use will differ.

Check a repository for this

OpenAI model IDs that shut down, fine-tunes on them, and calls that break on gpt-6-luna (tools, sampling, effort).

Runs in your browser with 2 anonymous GitHub API requests. Example: father-bot/chatgpt_telegram_bot@3631d6534f2c.

The same rules run offline, with a --fix for the safe rewrites, in atokim4/llm-retire-scan (MIT).

Sources

Dates and wording re-read on the official pages on . Vendors move dates; the official page wins.