OpenAI Retires GPT-4o as It Consolidates ChatGPT Models

News

1. IIT-Bombay, Columbia Team Up to Build AI Backbone for Indian Manufacturing

2. India Bets on Digital Public Infrastructure Model for Education AI

3. OpenAI Retires GPT-4o as It Consolidates ChatGPT Models

4. Adani Power Sets Up Nuclear Subsidiary

5. Musk Unveils xAI Overhaul, Lunar AI Ambitions

6. Former GitHub CEO Dohmke Raises $60 Million to Build AI Code Infrastructure

OpenAI is retiring several legacy models from ChatGPT, effective Friday, including GPT-4o, GPT-4.1, GPT-4.1 mini and o4-mini, alongside the previously announced sunset of GPT-5 Instant and GPT-5 Thinking.

While the API versions remain unchanged for now, the shift within ChatGPT marks the formal end of GPT-4o as a selectable model. 

While the retirement affects multiple models, GPT-4o has drawn outsized attention. It had already been briefly deprecated during the rollout of GPT-5 before being restored following strong user feedback. At the time, a section of Plus and Pro users made it clear they weren’t ready to move on.

Many said GPT-4o wasn’t just another model, it had a conversational warmth and fluidity they felt newer versions lacked. Others pointed to specific use cases, especially creative ideation and nuanced, emotionally intelligent conversations, where they believed GPT-4o performed better.

OpenAI says that feedback directly influenced the development of GPT-5.1 and GPT-5.2. According to the company, these newer versions incorporate stronger personality tuning, better creative support, and expanded customization.

Users can now select base styles and tones, such as “Friendly”, and adjust settings such as warmth and enthusiasm. The broader aim, OpenAI said, is to give users more control over how ChatGPT feels, not just what it can do.

The company said the transition reflects actual usage patterns, adding that the “vast majority” of users have already moved to GPT-5.2, with only 0.1% of daily users still selecting GPT-4o. With improvements now in place, OpenAI said it is ready to retire the older model.

On X, the reaction has been sharper.

Users rallied around the hashtag #keep4o, arguing the model functions as more than a preference. For some, it is an accessibility tool. 

“You are deprecating a vital accessibility aid,” one user wrote. 

Another said GPT-4o provides “vital support” for people with chronic illness, disabilities and neurodivergence, and that retiring the model “will cause global widespread harm.”

In a particularly detailed message, a user recovering from a chronic illness described GPT-4o as “a lifeline.” They explained that the model helps them manage energy levels, prepare questions for medical appointments, and navigate symptom flare-ups on difficult days. For them, GPT-4o’s tone and responsiveness were not cosmetic features but critical support mechanisms.

The responses highlight a growing tension in AI development: as companies iterate rapidly and optimize for scale, cost, and unified model architectures, small but highly dependent user segments may feel left behind.

Some users have even proposed alternative solutions. “You guys should just put 4o behind a 10k a month subscription. See how bad people need it,” one post read, suggesting that OpenAI could preserve access as a premium option rather than retire it outright.

OpenAI acknowledged that “losing access to GPT-4o will feel frustrating for some users” and said the decision was not made lightly. “Retiring models is never easy,” the company stated, but argued that doing so allows it to focus resources on improving the models most people use today.

Beyond the retirement, OpenAI also signaled broader changes underway. The company said it is continuing to refine ChatGPT’s personality and creativity while addressing user complaints about unnecessary refusals and overly cautious or “preachy” responses. Updates targeting those issues are expected soon.

In addition, OpenAI said it is working toward a version of ChatGPT “designed for adults over 18,” grounded in what it describes as treating adults like adults, while maintaining appropriate safeguards. As part of that shift, it has rolled out age prediction systems in most markets to identify users under 18.