OpenAI to Retire GPT-4o, GPT-4.1, and o4-Mini in ChatGPT

Farewell, ancient ones. OpenAI says it will remove several legacy AI models from ChatGPT.

On Feb. 13, 2026, the models GPT-4o, GPT-4.1, GPT-4.1 mini, and OpenAI o4-mini will be retired.

The change is limited to ChatGPT’s consumer and in-app model picker. “In the API, there are no changes at this time,” OpenAI said in the announcement, signaling that developers building on OpenAI’s platform won’t see immediate forced migrations as a direct result of this specific ChatGPT retirement announcement.

What’s changing in ChatGPT

OpenAI framed the February update as part of a broader model lineup refresh timed with “the previously announced retirement” of GPT-5 (Instant and Thinking) within ChatGPT.

In practical terms, users who still manually select older models for certain workflows—such as drafting, brainstorming, or maintaining a preferred conversational tone—will lose the ability to start new chats with those options once the retirement takes effect.

OpenAI did not provide detailed instructions in the announcement for how existing chats will behave after the switch, but the company emphasized it would “be clear about what’s changing and when.”

Special treatment

Although multiple models are being retired at once, OpenAI singled out GPT-4o for extra explanation, reflecting how sticky the model has been with a portion of the user base. The company said it had “first deprecated it and later restored access during the GPT-5 release,” and that the reversal was driven by feedback from Plus and Pro users.

Those users, OpenAI said, wanted more time to migrate “key use cases, like creative ideation,” and they preferred “GPT-4o’s conversational style and warmth.”

That framing underscores an emerging reality in consumer AI products: performance is not the only differentiator. Many users choose models based on feel—how collaborative, direct, or personable the assistant seems—especially for creative work and open-ended exploration.

Cover the gap

OpenAI argues the reason it can retire GPT-4o now is that the next-generation models have incorporated the very feedback that led to GPT-4o’s temporary return. The company said that feedback “directly shaped GPT-5.1 and GPT-5.2,” pointing to “improvements to personality, stronger support for creative ideation, and more ways to customize how ChatGPT responds.”

The broader implication is that OpenAI is trying to replace model-hopping with preference controls: rather than keeping older models around to satisfy distinct “vibes,” the company wants users to dial in behavior through settings, base styles, and tone controls like “Friendly,” plus adjustments for “warmth and enthusiasm.”

Logic of sunsetting

OpenAI says usage has already moved on. The company reported that “the vast majority of usage has shifted to GPT-5.2,” with “only 0.1% of users still choosing GPT-4o each day.”

That statistic matters because each additional model offered in a consumer product adds complexity—support burden, evaluation overhead, safety monitoring, and UI clutter. Retiring less-used options can also free up engineering attention and compute capacity for the models most people rely on.

Previews

OpenAI used the retirement notice to preview upcoming changes beyond model availability. The company said it is working on “further improvements to personality and creativity,” and also on reducing “unnecessary refusals and overly cautious or preachy responses,” with updates “coming soon.”

It also reiterated an effort to build “a version of ChatGPT designed for adults over 18,” grounded in the principle of “treating adults like adults,” while “expanding user choice and freedom within appropriate safeguards.”

To support differentiated experiences, OpenAI said it has rolled out “age prediction” intended to identify users under 18 in most markets.

Next steps

For most users, the immediate implication is simple: if you still depend on GPT-4o, GPT-4.1, GPT-4.1 mini, or o4-mini in ChatGPT—whether for writing voice, brainstorming style, or workflow familiarity—Feb. 13 is the deadline to adjust.

The UK is throwing the doors open to AI training.