Altman promises fixes after GPT-5 rollout glitches
OpenAI CEO Sam Altman and members of the GPT-5 team took questions during a Reddit “Ask Me Anything” (AMA) session on Friday, facing criticism over the new model and requests to restore GPT-4o, reports a Kazinform News Agency correspondent.
Altman admitted GPT-5 “seemed dumber” at launch because the real-time router that selects between models was not working properly. “GPT-5 will seem smarter starting today,” he said, adding that users will soon see more transparency about which model answers their queries.
Many participants urged the return of GPT-4o, which Altman said the company is considering for Plus subscribers. He also promised to double rate limits for paying users as the rollout continues.
The AMA also touched on the “chart crime” from the live presentation, where a misleading graph drew widespread ridicule. Altman later called it a “mega chart screwup.”
Altman closed by assuring users that fixes are underway: “We will continue to work to get things stable and will keep listening to feedback.”
Earlier, it was reported that OpenAI introduced a series of updates to ChatGPT designed to improve the model’s responses in emotionally complex situations, personal decision-making, and time management.