OpenAI has officially put a clock on one of its most beloved AI models. The company has announced that GPT-4o will be retired from ChatGPT on February 13, marking the end of a model many users still consider unmatched for writing nuance and emotional depth.
This isn’t GPT-4o’s first brush with retirement. Back in August 2025, OpenAI attempted to sunset the model after rolling out GPT-5, only to reverse the decision following strong user backlash. At the time, users argued that newer models felt more mechanical and lacked GPT-4o’s natural writing flow. This time, however, OpenAI appears firm.
Which Models Are Being Retired
Alongside GPT-4o, OpenAI confirmed that several other legacy models will also be removed from ChatGPT, including:
- GPT-5 Instant
- GPT-5 Thinking
- GPT-4.1
- GPT-4.1 mini
- o4-mini
Importantly, this change only affects ChatGPT. Developers using OpenAI’s API will continue to have access to these models, at least for now.
Why OpenAI Says GPT-4o Is No Longer Needed
In its announcement, OpenAI gave special attention to GPT-4o—likely because of the backlash it faced previously. The company says it has spent months studying how people actually used the model and has since transferred those strengths into GPT-5.2.
According to OpenAI, newer models now match GPT-4o in:
- Creative ideation
- Writing personality
- Emotional tone
- Customisation and style control
Users can now tweak response styles, adjust warmth and enthusiasm, and choose preset base personalities—features OpenAI claims were inspired directly by GPT-4o’s popularity.
The company also stated that only 0.1% of users still select GPT-4o daily, suggesting the audience holding on to it has become extremely small.
Users Aren’t Buying the Numbers
That 0.1% figure hasn’t gone down well online.
Several users on X (formerly Twitter) have accused OpenAI of presenting a misleading picture. One user argued that once GPT-5.2 became the default, most users simply stopped manually switching models—especially since GPT-4o requires a paid subscription and is sometimes interrupted by safety routing.
Others pointed out that thousands of posts protesting the decision were trending within hours, questioning how a model allegedly used by “almost no one” could spark such a reaction.
Even OpenAI Admits Writing Took a Hit
Adding fuel to the fire, Sam Altman recently acknowledged during OpenAI’s first company-wide town hall that writing quality declined in GPT-5.2. According to Altman, the shift happened because the company prioritised coding, reasoning, and mathematics during training.
That admission has only strengthened the argument of users who feel GPT-4o represented a creative high point that newer models haven’t fully recaptured.
What This Means Going Forward
For OpenAI, this move signals a clear direction: consolidation around newer models, even if it means upsetting a vocal minority. For users—especially writers, creatives, and long-form storytellers—it feels like losing a trusted collaborator.
Final Words
GPT-4o’s retirement isn’t just about sunsetting an old model—it highlights a deeper tension in AI development: technical power versus human-like expression. While OpenAI insists the future models can do it all, many users remain unconvinced.
Come February 13, one of ChatGPT’s most emotionally resonant voices will go quiet. Whether GPT-5.2 truly fills that gap—or drives more users away—will become clear soon enough.
