OpenAI Restores GPT-4o in ChatGPT and Adjusts GPT-5 Usage Limits

OpenAI Enables Paid Users to Access GPT-4 Directly in ChatGPT Model Picker

OpenAI has reinstated the GPT-4o AI model in ChatGPT following user feedback. The model had been retired last week after the launch of GPT-5, which the company positioned as the new flagship capable of advanced reasoning and conversational tasks. Many users expressed a preference for GPT-4o’s responses over GPT-5 on social media, prompting the San Francisco-based AI firm to make it available once again for all paid subscribers.

Sam Altman, CEO of OpenAI, announced the updates on X (formerly Twitter), highlighting several changes specifically for paid users, including ChatGPT Plus, Team, and Pro subscribers. The key update is the return of GPT-4o, which will remain accessible indefinitely, though the company noted it will provide advance notice if it decides to retire the model in the future.

Alongside GPT-4o, OpenAI is also reintroducing older models such as o3, GPT-4.1, and GPT-4.5. The GPT-4.5 model will be exclusive to Pro users due to its high computational requirements. Users can enable these models directly through ChatGPT’s web settings, offering flexibility to choose the AI version best suited for their tasks.

Additionally, paid subscribers now have access to GPT-5 Thinking Mini, a lighter variant of GPT-5 that can handle queries with reduced computational load. These updates collectively aim to give users more options for balancing performance, cost, and response style across OpenAI’s AI models.