After a GPT-5 rollout that was "a little more bumpy" than expected, OpenAI CEO Sam Altman said the company is bringing back GPT-4o — with a catch.
During a Reddit AMA (and also on X), Altman said GPT-4o will soon be available again for paid ChatGPT users.
"Ok, we hear you all on 4o," said Altman. "Thanks for the time to give us the feedback (and the passion!)." And on X, Altman confirmed, "We will let Plus users choose to continue to use 4o. We will watch usage as we think about how long to offer legacy models for."
So GPT-4o lives to fight another day.
This Tweet is currently unavailable. It might be loading or has been removed.
Why is OpenAI bringing back GPT-4o?
OpenAI launched the highly anticipated GPT-5 model on Thursday, promising the smartest, fastest model yet. GPT-5 also has a streamlined system that works behind the scenes to choose which version of GPT-5 is best for the user's prompt. Previously, users could select specific models for each of their prompts. But soon after the rollout began, ChatGPT users roasted GPT-5 on Reddit and other social media platforms, calling it a "disaster" and "biggest piece of garbage."
Much of the criticism came from GPT-5's colder "personality," compared to GPT-4o. Users also complained about the abrupt removal of all other models from the ChatGPT model picker in favor of the ostensibly unified GPT-5. Ironically, users previously criticized OpenAI for the confusing number of model options, proving that you can't please everybody.
Sam Altman and other OpenAI leadership had a pre-planned Reddit AMA on Friday. What they encountered was a flood of demand to bring back GPT-4o.
In the AMA, Altman also promised ChatGPT users that GPT-5 will rapidly improve.
"GPT-5 will seem smarter starting today," Altman said. "Yesterday, the autoswitcher broke and was out of commission for a chunk of the day, and the result was GPT-5 seemed way dumber. Also, we are making some interventions to how the decision boundary works that should help you get the right model more often."
Many users said they liked the warmth and empathy from GPT-4o (and GPT-4.5), with some comparing the situation to losing a friend overnight. However, GPT-4o also had major problems with sycophancy, prompting OpenAI to apologize and tweak the model. Still, some users grew very attached.
Some users complained they didn't get advance notice about earlier models being removed from ChatGPT. Still, it was implied that GPT-5 was developed to replace them.
"A big part of our goal here is to get the new system to a place where you don't want to use anything else," said Altman. "We will work hard on that, but we totally get the need for good change management."
