During a Reddit ask-me-anything session on Friday, OpenAI CEO Sam Altman and key members of the GPT-5 team answered questions about the new model. Many users requested the return of the previous model, GPT-4o, and also brought up the amusing “chart crime” mishap from the presentation.
One of GPT-5’s new features is a real-time router that selects the best model for each prompt, either responding quickly or taking more time to process complex answers. However, several users in the AMA complained that GPT-5 wasn’t performing as well as GPT-4o. Altman explained that the issue stemmed from a malfunctioning router when the model was first launched.
He assured users that improvements were being made, stating, “GPT-5 will seem smarter starting today. Yesterday, we had a system issue, and the autoswitcher was down for part of the day, which made GPT-5 appear less capable. We’re also adjusting how the decision boundary works to ensure the right model is chosen more often. We’ll make it clearer which model is handling each query.”
Despite these fixes, many users pushed for GPT-4o to remain available for Plus subscribers. Altman acknowledged the request, saying, “We are looking into letting Plus users continue to use GPT-4o. We need more data on the tradeoffs.” He also announced plans to double the rate limits for Plus users, giving them more flexibility to explore the new model without exhausting their monthly prompts.
The “chart crime” from the presentation became a popular topic during the AMA. The chart in question displayed incorrect proportions, with a lower benchmark score appearing as a taller bar. Altman didn’t address it directly in the session but had previously called it a “mega chart screwup” in a post. Corrected versions were later published in OpenAI’s official blog post, but the mistake had already sparked jokes about relying on AI for corporate presentations.
Early reviewer Simon Willison, who generally praised GPT-5, noted that converting data into tables remained a weak point for the model. Altman closed the AMA by promising further improvements, saying, “We will continue to work on stability and keep listening to feedback.”
The TechCrunch Disrupt 2025 event was also highlighted, featuring industry leaders like Netflix, ElevenLabs, Wayve, Sequoia Capital, and Elad Gil. The conference, taking place in San Francisco from October 27-29, 2025, aims to provide insights for startup growth and innovation. Attendees were encouraged to register early for discounted tickets.