Sam Altman accepts mistakes with GPT-5 launch

Sam Altman, the CEO of OpenAI, has openly recognized the missteps made during the launch of the GPT-5 model. Following the introduction of this latest AI iteration, OpenAI faced significant backlash after discontinuing all previous models, prompting many users to threaten to cancel their subscriptions. In response to the criticism, OpenAI quickly reverted some decisions, reintroducing the GPT-4o model for ChatGPT Plus users and adjusting rate limits for both the GPT-5 Standard and Thinking models. While the company touted GPT-5 as a revolutionary advancement in areas such as coding, reasoning, and multimodal capabilities, many users reported dissatisfaction with the new model. They noted that GPT-5's responses were often shorter and lacked the emotional depth that had been characteristic of earlier versions. Altman, who had been promoting GPT-5 as a monumental step forward in AI, seemed taken aback by the negative feedback. In a recent interview, he candidly stated, “I think we totally screwed up some things on the rollout.” He acknowledged the challenges of upgrading a product that serves hundreds of millions of users and the unique attachment they have to it compared to earlier products. Despite the rocky start, Altman revealed that traffic to the ChatGPT API doubled in the 48 hours following the GPT-5 release, and app usage reached unprecedented levels. He admitted that discontinuing older models was a miscalculation on their part. In the week leading up to the launch, OpenAI celebrated reaching a milestone of 700 million weekly users, a figure the company hoped to expand upon with the introduction of new models. Unlike the previous major launch that had sparked a worldwide craze for its image generation capabilities, the response to GPT-5 was markedly different, with social media buzzing with criticism instead of excitement. This backlash even led OpenAI to reinstate the model picker, a feature Altman had promised to remove in the lead-up to the launch.