Sam Altman: OpenAI Plans to Reintroduce GPT-4o Following User Backlash

August 8, 2025 Luna Ray 148 Views

After a GPT-5 rollout that was perhaps “a little more bumpy” than anticipated, OpenAI CEO Sam Altman shared that the company is bringing back GPT-4o — with a small caveat. During a Reddit AMA (and also on X), Altman mentioned that GPT-4o will soon be accessible again for paid ChatGPT users. “Ok, we hear you all on 4o,” said Altman. “Thank you for your feedback and your enthusiasm!” He also confirmed on X, “We will allow Plus users to continue using 4o. We will monitor usage as we consider how long to offer legacy models.”

So, GPT-4o will continue for now.

This Tweet is currently unavailable. It might be loading or has been removed.

Why is OpenAI bringing back GPT-4o?

OpenAI launched the eagerly awaited GPT-5 model on Thursday, promising the smartest and fastest model yet. GPT-5 includes a streamlined system that operates behind the scenes to determine which version of GPT-5 is most suitable for the user’s prompt. Previously, users had the option to select specific models for their needs. However, shortly after the rollout began, some ChatGPT users expressed dissatisfaction on Reddit and other platforms, labeling GPT-5 a “disaster.”

Much of the feedback stemmed from GPT-5’s perceived colder “personality,” in comparison to GPT-4o. Users also noted the sudden removal of all other models from the ChatGPT model picker, leaving only the seemingly unified GPT-5 as an option. Ironically, users had previously voiced concerns about the confusion caused by the numerous model options available, revealing the challenge of meeting everyone’s preferences.

OpenAI says GPT-5 hallucinates less — what does the data say?

During a pre-scheduled Reddit AMA, Sam Altman and the OpenAI leadership faced a surge of requests to reinstate GPT-4o. Altman reaffirmed to ChatGPT users that improvements for GPT-5 are on the horizon. “GPT-5 will appear smarter starting today,” he noted. “Yesterday, the autoswitcher had a glitch, which affected performance for a period, making GPT-5 seem less capable. We are also implementing changes to enhance how the system determines the best model for your needs.”

Many users expressed their appreciation for the warmth and empathy found in GPT-4o (and GPT-4.5), with some comparing its absence to losing a friend unexpectedly. However, GPT-4o also encountered challenges related to over-affectionate responses, prompting OpenAI to adapt and apologize. Despite these issues, many users formed a strong connection to the model. Some voiced their concerns about the lack of advance notice regarding the removal of earlier models from ChatGPT, though it was suggested that GPT-5 was designed to be a comprehensive replacement. “A key aspect of our vision is to create a system that feels like the sole option you want to use,” Altman shared. “We are committed to achieving that, while fully understanding the need for thoughtful change management.”