(Bloomberg) -Sam Altman, CEO of Openai, has been setting up the capabilities of GPT-5 for months and preparing the launch as an important moment for the company. But in the first 24 hours after its release, the new model was met with mixed reviews. In his announcement on Thursday, Openai said GPT-5 is better in coding and reasoning through complicated problems, and has designated it as advanced enough to make chatbot chatgpt in a Ph.D. level. Some with early access priced the model with warnings. “This is my new favorite model,” wrote the developer Simon Willey in a blog post and called it “competent” and “sometimes impressive”. He added: “This is not a dramatic departure from what we had before.” However, on various social media platforms, Chatgpt users expressed frustration that GPT-5 was still setting up information and was about simple math and game questions. Noah Giansiracusa, an associate professor of mathematics at Bentley University, said he felt the launch was ‘under -waring’. While there were some improvements, he said, “they were much more marginal than I would have hoped.” At least some of the reaction may amount to confusion about what happens under the bonnet. Unlike Openai’s previous software, GPT-5 switches automatically between models of different levels of sophistication, depending on the query. This approach can help maximize the computer sources of the business, but it also means that users are not always involved with the most powerful version of Openai’s technology. Caught to identify how many times the letter “B” appears in “Blueberry”, for example, GPT-5 initially said “three” in one test. When he told ‘to think harder’, GPT-5 seemed to involve his more advanced reasoning model and the right answer came up. Altman responded to some of the feedback on Friday, saying that there was a problem with the system. “GPT-5 will look smarter today,” he said. “Yesterday the Autoswitcher broke and he was out of order for part of the day, and the result was that GPT-5 was very dumber.” The interests are high for the launch. Openai fights to continue the growth of AI competition of opponents in the US and China. The company is also fighting to convince businesses and individual users to pay for its premium services to compensate the enormous amount it spends on talent, chips and data centers to support AI development. The company in San Francisco kicked off the generative AI tree almost three years ago with the release of Chatgpt, originally powered by an earlier model called GPT-3.5. Since then, the company has released a series of increasingly sophisticated systems, including various options that mimic the process of human reasoning. When AI systems progress, it has become harder to definitely say how different services stand up. From noon Friday, GPT-5 rose to the top of different categories on LMARENA, a popular AI models scoring list based on user rankings. But another benchmark, ARC-AGI-2, places GPT-5 behind the latest version of Grok of Elon Musk’s Xai. In the absence of more definite assessments, the model wars sometimes come across vibes. And with nearly 700 million people now using chatgpt, some will not agree on how the model feels. It also takes longer than a day to determine the value of a new AI system in someone’s personal and professional life. Ethan Mollick, a professor at the Wharton School of the University of Pennsylvania who regularly experimented with AI models, marvel at GPT-5’s ability to do research, come up with smart written answers and make programming, even for a beginner. “GPT-5 just does things, often extraordinary things, sometimes strange things, sometimes a lot of ai-things on his own,” he wrote in a blog post. “And that’s what makes it so interesting.” However, on Reddit the reactions were very different. During an “Ask Me Anything” session on the platform on Friday, Altman set back from users who were frustrated to no longer say and have visibility in which model respond to their questions. Altman said Openai would take a few steps to address these complaints, including it ‘more transparent’. At one point, Altman responded to the question of a Reddit user by noting that Openai thinks that the ‘writing quality’ in one version of GPT-5 is better than GPT-4.5. Then he asked, “Do you find it worse?” One user after another was quick to respond: Yes. More stories like these are available on Bloomberg.com © 2025 Bloomberg LP
Openai’s GPT-5 met mixed reviews, confusion on the first day
