OpenAI recently launched GPT-5, touted as their most powerful model yet, but it quickly faced backlash from users over its performance and changes in user experience. In response, CEO Sam Altman announced the reinstatement of GPT-4o for ChatGPT Plus subscribers while the company examines user feedback.
GPT-5 aimed to unify previous models, employing a real-time routing system designed to optimally select engines for each prompt. However, users encountered issues, partly due to an outage affecting this autoswitcher, which fueled perceptions of decreased capability. Altman acknowledged these problems and promised adjustments for better transparency regarding which model handles specific requests.
User dissatisfaction wasn’t just about performance. Many valued the warmer, more conversational tone of GPT-4o, and found GPT-5 colder and more mechanical. This shift sparked debates across platforms like Reddit, with some users threatening to cancel subscriptions if GPT-4o wasn’t reinstated. OpenAI maintains that the availability of GPT-4o will depend on user demand, and they are closely monitoring usage patterns.
The rollout faced additional challenges due to a charting error, which misrepresented benchmark results. This was corrected in OpenAI’s blog with updated and accurate information.
During a Reddit AMA, Altman reassured users that GPT-5 will improve as the routing system stabilizes and reiterated the company’s commitment to user feedback. OpenAI is considering allowing Plus users the option to continue with GPT-4o while analyzing the trade-offs between the two models.
Altman has made it clear they’re listening and adapting based on user input, and OpenAI plans to observe trends before making further decisions about the long-term support of GPT-4o.
