OpenAI CEO Sam Altman and members of the GPT-5 team answered user questions during a Reddit Ask-Me-Anything (AMA) session. Users shared their experiences with the new GPT-5 model, with some expressing a desire to bring back the previous GPT-4o version.
Altman highlighted a new feature of GPT-5—a real-time “router” system that chooses which model to use for each prompt. This system aims to provide quick responses for some queries while taking extra time to “think” through others. However, during the rollout, the router malfunctioned, causing GPT-5 to seem less capable than its predecessor.
“GPT-5 will perform smarter starting today. Yesterday, there was a serious issue and the auto-switching feature was down for a while, which made GPT-5 seem weaker. We’ll also make it clearer which model is responding,” Altman explained.
When users requested that GPT-4o remain available to Plus subscribers, Altman said they are considering the possibility: “We are looking into letting Plus users continue to use GPT-4o and are gathering more data.”
Additionally, Altman promised to double the monthly usage limits for Plus subscribers to help users comfortably test and adapt to the new model.
The session also covered the widely criticized “chart error” from GPT-5’s live presentation. Altman acknowledged on Twitter that the chart was a “major screw-up.” The presentation showed a lower benchmark score with a taller bar, which confused many viewers. However, the official blog post contained the correct charts.
This mistake sparked jokes on social media and criticism over using GPT to create charts for corporate presentations. Early GPT-5 reviewer Simon Willison pointed out the model’s failure to properly convert data into tables.
Altman concluded the AMA by assuring users that OpenAI will continue working to stabilize the system and is committed to listening to user feedback and suggestions.
