Sam Altman, CEO of OpenAI, directly addressed the performance controversy raised immediately after the release of the latest artificial intelligence (AI) model 'GPT-5.'
On the 8th (local time), Altman participated in an 'Ask Me Anything' (AMA) session on the U.S. community-based social media platform Reddit, where he answered user questions alongside the GPT-5 development team. During the AMA session, criticism poured in that 'GPT-5 performs worse than its transfer model, GPT-4o.' Some users requested, 'Please allow us to access GPT-4o again.'
Altman explained that the reason GPT-5 appeared to have degraded performance was that the 'router,' which determines in real-time which model to use based on user prompts (instructions), did not operate properly at the time of the new model's launch on the 7th. This new feature added to GPT-5 determines whether a quick response is needed or if more time should be taken for a deeper answer, providing the best response accordingly. Altman said, 'On the first day, the model auto-selection feature did not function due to a serious error, making GPT-5 appear dumber. In the future, we will adjust the model selection criteria to support users in utilizing the most suitable model, and we will make it more transparent to show which model is currently responding to the question.'
Some Plus subscribers among the AMA participants requested to be allowed to use GPT-4o again. Altman noted, 'We are reviewing the possibility for Plus subscribers to continue using 4o,' adding, 'We are trying to gather more data to compare the pros and cons.'
He also stated that once GPT-5 implementation is complete, he would double the usage limits for Plus subscribers.
He mentioned, 'The deployment of GPT-5 for all ChatGPT users is taking longer than expected,' noting that 'API traffic nearly doubled in just the past 24 hours.' The weekly active users of ChatGPT reach 700 million.