OpenAI is retiring several older ChatGPT models on February 13, 2026, as part of its ongoing transition to newer AI technology. The models being phased out include GPT-4o, GPT-4.1, GPT-4.1 mini, and OpenAI o4-mini, following earlier announcements regarding the retirement of GPT-5 Instant and Thinking.
The decision shows how user feedback is increasingly influencing OpenAI’s model lifecycle, particularly with GPT-4o. After the model was initially deprecated during last year’s GPT-5 launch, OpenAI reversed course and restored access in response to strong feedback from some Plus and Pro subscribers. Many cited the need for additional time to adapt existing workflows, especially around creative ideation, and pointed to GPT-4o’s conversational tone as a key differentiator.
OpenAI says user feedback helped guide the development of its latest flagship models, GPT-5.1 and GPT-5.2. The updates include refinements to personality and creative ideation features, as well as more detailed response customization. Users can now select base styles, such as “Friendly,” and adjust attributes like warmth and enthusiasm, reflecting a focus on interaction style in addition to core functionality.
The announcement cites the successful integration of these features and a significant change in user adoption as the main reasons for the retirement. OpenAI reports that GPT-4o is now used by just 0.1% of daily users, with most activity concentrated on GPT-5.2.
The API (Application Programming Interface) for developers remains unaffected by this change at this time.
More broadly, OpenAI says it is continuing to refine ChatGPT in response to recurring user criticism. Areas of focus include further improvements to personality and creativity, along with efforts to cut back on what the company describes as unnecessary refusals and overly cautious or preachy responses. Additional updates are expected soon.
OpenAI says it recognizes the disruption the change may cause and that the early announcement is meant to allow time for adjustment. In the notice, the company says it understands that losing access to GPT-4o will be frustrating for some users and emphasizes that retiring older models frees up engineering and computing resources to improve the models used by most users today.
Maybe you would like other interesting articles?