OpenAI to Sunset GPT-4o, GPT-4.1, and o4-Mini in ChatGPT Model Refresh
Table of Contents
OpenAI is streamlining its ChatGPT offerings, announcing the retirement of several legacy models – GPT-4o, GPT-4.1, GPT-4.1 mini, and o4-mini – on February 13, 2026. This move signals a broader strategy to consolidate resources and focus on its latest generation of AI, while also acknowledging the surprisingly strong emotional connection some users have developed with older iterations.
A Shift in ChatGPT’s Core Models
The upcoming changes will impact ChatGPT’s consumer and in-app model picker, but will not immediately affect developers utilizing OpenAI’s API. According to a company release, developers building on the OpenAI platform will not face forced migrations as a direct result of this specific update. The retirement is framed as part of a larger refresh, coinciding with the previously announced sunsetting of GPT-5 (Instant and Thinking) within ChatGPT.
For users who have grown accustomed to manually selecting specific models for tasks like drafting, brainstorming, or achieving a particular conversational tone, the February 13th deadline marks a turning point. Once the retirement takes effect, the ability to initiate new chats with these older options will be lost. OpenAI has stated it “will be clear about what’s changing and when,” but has not yet detailed how existing chats will be affected.
The Unexpected Stickiness of GPT-4o
While multiple models are being retired, OpenAI devoted extra attention to explaining the decision regarding GPT-4o. The company revealed it had initially deprecated the model, only to restore access during the release of GPT-5 due to user feedback.
“Plus and Pro users wanted more time to migrate key use cases, like creative ideation,” one official stated, “and they preferred GPT-4o’s conversational style and warmth.” This highlights a growing trend in consumer AI: performance isn’t the sole determinant of user preference. Many users prioritize the feel of an AI assistant – its collaborative nature, directness, or personality – particularly for creative endeavors.
From Model-Hopping to Preference Controls
OpenAI believes it can now retire GPT-4o because subsequent models have incorporated the feedback that drove its temporary return. The company asserts that user input “directly shaped GPT-5.1 and GPT-5.2,” resulting in “improvements to personality, stronger support for creative ideation, and more ways to customize how ChatGPT responds.”
This suggests a shift away from offering a multitude of distinct models and toward a system of granular preference controls. Rather than maintaining older models to cater to specific “vibes,” OpenAI aims to allow users to fine-tune behavior through settings, base styles, and tone adjustments like “Friendly,” with further controls for “warmth and enthusiasm.”
Usage Trends and the Logic of Sunsetting
Data released by OpenAI indicates that the vast majority of users have already transitioned to GPT-5.2, with a remarkably low 0.1% still choosing GPT-4o daily. This statistic underscores the practical rationale behind the move. Maintaining multiple models adds complexity in terms of support, evaluation, safety monitoring, and user interface design. Retiring less-used options frees up engineering resources and computing power for the models most widely adopted.
Future Enhancements and Adult-Focused AI
The retirement announcement also served as a preview of upcoming improvements. OpenAI is actively working on “further improvements to personality and creativity,” as well as reducing “unnecessary refusals and overly cautious or preachy responses.” Updates are expected “coming soon.”
Furthermore, OpenAI reiterated its commitment to developing a ChatGPT version specifically designed for users over 18, emphasizing “treating adults like adults” while “expanding user choice and freedom within appropriate safeguards.” To facilitate this, the company has implemented “age prediction” technology to identify users under 18 in most markets.
For the vast majority of users, the immediate takeaway is clear: if you currently rely on GPT-4o, GPT-4.1, GPT-4.1 mini, or o4-mini in ChatGPT – whether for a specific writing style, brainstorming approach, or established workflow – February 13, 2026, is the deadline to adapt.
