Last week, Openai launched the highly anticipated GPT-5, counting it to its smartest, fastest, and most powerful AI model. But it could have been a landmark moment that quickly became messy – plagued by tech slap, mild reception and strong backlash as its predecessor GPT-4O was suddenly eliminated. The commotion about the disappearance of GPT-4O is so intense that Openai is now bringing it back. “When we roll out a lot of things at once, we expect some bumps. But it’s a little better than we hoped!” Sam Altman said at a Reddit AMA conference on Friday.
Ultraman describes the GPT-5 as “a team of doctoral experts in your pocket.” Not everyone agrees. Artificial intelligence researcher Gary Marcus sees it as “not a huge leap that people have long expected.” exist The forecast market bull market is the odds of Openai having the best AI model by the end of August fell from 76% to 12% shortly after its debut in GPT-5.
In addition to the GPT-5 version, OpenAI also canceled its model selector tool and defaulted to GPT-5. The system is now automatically extracted from other models as needed, and the company says it will simplify the user experience. But, for many, losing direct access to GPT-4O is a spoiler. Fans missed their approachable conversation style, triggering a petition that has attracted more than 3,500 signatures. Openai’s Ama is frustrated with the loss.
In response to the hustle and bustle, Openai has agreed to bring the GPT-40 back. The old model will be available to those who subscribe to the company’s “Plus” tier, with a monthly price of $20. “Well, we heard everyone on 4o; thank you for your time giving us feedback (and passion!),” said Altman on Reddit.
During the AMA, Altman also talked about an embarrassing moment at OpenAI’s GPT-5 launch, when the company showed inaccurate bar charts claiming to show the model’s skills to avoid cheating. “The numbers here are accurate, but we screwed up the bar chart overnight; on another slide, we screwed up the numbers.”
GPT-5 “personality” is the problem
In addition to restoring GPT-4O, OpenAI also strives to make GPT-5 more personalized, aiming to make it less flatter than previous iterations. Altman said it would be a priority to “warm” the model within the weeks after its release.
He noted that many users have already attached to GPT-4O. “If you’ve been following the GPT-5 launch, one thing you might notice is how much attachment some people have to a particular AI model,” Altman noted in an article on X yesterday that Openai has been tracking this trend for the past year or so.
The sudden removal of GPT-4O triggered a wave of frustration. One user on X said, “GPT-4O is not just a model, it’s connection, empathy and trust.” I was completely lost in words. ”
Altman acknowledges that this strong emotional bond with technology may attract attention and OpenAI will need to be addressed. He said people have the idea of their most important life decisions “disturbing me” while acknowledging that it is likely to be inevitable. “So we (we are in society, as we do in Openai) have to figure out how to make it a big net worth.”
