OpenAI confirmed today that it will retire several legacy models, including GPT-4o, from the consumer-facing ChatGPT interface starting February 13, 2026. This date also marks the official retirement of the previously announced GPT-5 (Instant and Thinking) variants. Currently, the API infrastructure for these models remains unchanged, according to the announcement made on OpenAI's official blog.
The context surrounding GPT-4o's removal is notable, as the model was temporarily deprecated and then restored following strong user demand. Certain Plus and Pro subscribers cited the need for more time to migrate critical use cases, specifically praising GPT-4o’s conversational warmth and creative ideation capabilities.
This direct user feedback reportedly served as a foundational data point for developing the subsequent iterations, GPT-5.1 and GPT-5.2. Those newer models incorporate refined personality traits, enhanced creative support, and expanded customization options, such as base styles like 'Friendly' and controls for enthusiasm levels.
OpenAI stated that the vast majority of user traffic has already migrated to GPT-5.2, with GPT-4o utilization dropping to only 0.1% of daily interactions. Retiring the older models allows the engineering teams to concentrate resources on optimizing the current, widely adopted versions, as reported by OpenAI.
Beyond model sunsetting, the organization continues to address user concerns regarding model behavior. Future updates aim to improve personality and creativity while mitigating instances of unnecessary refusals or overly cautious responses. This commitment reflects an ongoing effort to treat adult users with greater freedom within established safety parameters.
To support its commitment to treating adults like adults, OpenAI has begun rolling out age prediction capabilities for users under eighteen across most markets globally. The company acknowledged that model transitions can cause friction for some users but emphasized the necessity of sunsetting older iterations for platform efficiency.
These ongoing adjustments signal a strategic focus on refining the user experience through conversational nuance and control, moving beyond raw capability metrics. The industry will monitor how effectively GPT-5.2 incorporates the lessons learned from the GPT-4o user base to define the next generation of conversational AI interaction.
