OpenAI announced that on February 13, 2026 it will retire several older ChatGPT models, including GPT-4o, GPT-4.1, GPT-4.1 mini and OpenAI o4-mini. If you use the API you won’t see changes for now, but in the ChatGPT interface you’ll watch these models disappear.
What changes and when
The retirement date is February 13, 2026. This joins a previous cleanup that also removed versions of GPT-5 (Instant and Thinking). In practice, most users already migrated to GPT-5.2, and OpenAI says only 0.1% of users still choose GPT-4o each day.
What does this mean for you? If you open ChatGPT and were using one of those models, you’ll need to pick alternatives before or after the date. If you work with the API, OpenAI clarifies there are no changes at the moment.
Why these models are being retired
It’s not just cleanup for the sake of cleaning. GPT-4o was especially interesting: they retired it, brought it back after feedback, and then learned how people were actually using it. Some users preferred its conversational style and warmth for tasks like creative ideation.
Those lessons were folded into GPT-5.1 and GPT-5.2, with improvements in personality, creativity and customization options (base styles like Friendly and controls for warmth and enthusiasm). The idea is that the model not only does what you ask, but you can also control how the interaction feels.
What happens to your conversations and workflows?
If you rely on GPT-4o for creative notes, brainstorming or answers with a certain tone, you’ll have to adapt prompts and test how GPT-5.2 behaves. Worried about losing history or integrations? Practical recommendations:
- Export important conversations before the retirement.
- Test your prompts on
GPT-5.2and tweak the tone using the style controls. - If you use the API, verify your endpoints still work; for now there are no changes.
A concrete example: if you used GPT-4o for ideation sessions with a warm tone, try the Friendly style in GPT-5.2 and raise the warmth control slightly. You’ll likely find the feel is very similar or even better.
Alternatives and migration steps
- Migrating to
GPT-5.2is the main recommendation: it includes the personality and creativity improvements. - If you need time to test, save your prompts and create a verification plan (a test of 10–20 key prompts).
- For teams: communicate the change, assign someone to validate integrations and update internal docs.
Impact on policy and user experience
OpenAI also says it’s still working to reduce unnecessary rejections and overly cautious or lecturing responses. They’re moving toward a ChatGPT version aimed at users 18+ with more freedom inside proper safeguards. To support that, they’ve rolled out an age-prediction tool for users under 18 in most markets.
Be realistic: changes like this frustrate people who relied on a specific behavior, and OpenAI acknowledges that. Retiring models lets them focus resources on improving the models most people use, but it also forces adaptation.
Final reflection
If you’re a creator, professional, or everyday user, this is a call to try and adjust. The good news is the lessons from GPT-4o didn’t vanish: they’re inside the new versions and there are controls so the experience is more customizable. The bad news? There’s migration work if your flow depended on GPT-4o’s unique flavor.
In short: prepare your prompts, export the critical stuff and try GPT-5.2. The switch can be a short-term annoyance, but the platform aims to give you more control over how the conversation feels.