OpenAI gave GPT-4o API users three months' notice — its rule says six
Two deprecation notices landed in the same week, and they point in opposite directions: a consumer model pulled off ChatGPT on schedule, and five production API models given half the notice OpenAI normally promises.
OpenAI notified developers on October 14 that five dated GPT-4 API models shut down on January 14, 2027 — three months out, not the six its own deprecation policy requires for generally available models. The list: two GPT-4o snapshots (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. Fine-tuned versions of all five share the deadline — extra training time does not buy extra runway. OpenAI's recommended replacement is gpt-6-luna, which is a real migration, not a drop-in: the guidance says to start with reasoning effort set to none and test against representative traffic before shifting production.
The policy gap is the story. OpenAI's own deprecations page states that generally available models get "at least 6 months" notice, with a carve-out only when safety or compliance concerns require a faster timeline. This notice does not invoke that carve-out. Compare the same page's October 1 entry, where GPT-5.3-Codex, GPT-5.4-Nano and GPT-5.1 were given six months to April 1, 2027 — same company, same week, different clock.
That asymmetry matters more than the individual models. GPT-4o remains one of the most widely deployed model families in production software, and three months covers a holiday quarter in most engineering orgs: enough to update config strings, not always enough to revalidate prompts, retune reasoning settings and re-run regression suites under a model with different latency and token economics. The teams most exposed are the ones with the least slack — the long tail of small products that picked gpt-4o-mini in 2024 because it was cheap and never had a reason to revisit it.
The same week, GPT-5.5 retired from ChatGPT, ChatGPT Work and Codex on all plans — consumer, Business, Enterprise and Edu — effective October 14 at 12:00 PDT. That retirement explicitly did not touch the API. But it did force migration for anyone with gpt-5.5 baked into workspace defaults, saved model settings, custom agents, scheduled tasks or scripts: Plus, Pro, Business, Enterprise and Edu move to gpt-6.1-sol, Free and Go to gpt-6-luna. Two different notices, one week apart, and the model that stayed available was the one inside OpenAI's subscription.
Read the two together and the pattern is routine housekeeping, not conspiracy — end-of-life notices are how any platform manages a product line, and OpenAI has retired models on schedule all year. The defensible complaint is narrower: when the vendor's published commitment says six months and the notice says three, the commitment stops being a planning input. We flagged that API-sunset arithmetic once before, when GPT-4o's first API snapshot shut down on October 23 — the difference now is that the whole family is on the board at once, on the short clock.
What to watch: whether OpenAI amends the notice window or explains the shorter one. A pinned migration deadline with a published reason is a normal platform decision; a pinned deadline against your own published policy is the thing teams will plan around next time.
If a vendor's published notice period isn't binding, what would you actually plan against? Tell us in the comments.
