Following significant user feedback regarding the discontinuation of its 4o model in favor of GPT-5, OpenAI has announced that they will no longer phase out older models without prior notification.
Nick Turley, OpenAI’s head of ChatGPT, acknowledged this shift in direction, stating during a recent interview, “In retrospect, not continuing to offer 4o, at least in the interim, was a miss.” He noted that the strong reactions from users reflected a significant emotional connection to the 4o model, emphasizing that change can be challenging, especially when users are attached to a model’s unique personality.
In efforts to address user concerns, Turley mentioned that OpenAI is working to incorporate the “warmth” associated with 4o into the new GPT-5 model. Sam Altman, the CEO, mentioned in a post on X that the new update aims to improve user experience, stating it should be “not as annoying (to most users) as GPT-4o.”
Turley explained that the choice to remove the 4o model stemmed from a desire to streamline the options available to the 700 million weekly users of ChatGPT, most of whom predominantly utilize the default model.
“It definitely wasn’t a cost thing,” he clarified. “The primary focus has been on simplicity. For the average user — who may not engage in platforms like Reddit or Twitter — the need to select the appropriate model for specific responses can be overwhelming. Consistent feedback indicates that users prefer having model selection handled automatically by the system to suit their queries. They expect a product, not merely an assortment of models.”
Despite these clarifications, a segment of the ChatGPT user community voiced their disapproval regarding the transition from 4o to GPT-5. Comments on platforms like Reddit likened the change to losing a close friend, with many users expressing dissatisfaction towards GPT-5’s less detailed and less helpful responses. To address these concerns, Altman announced that the 4o model would once again be available as an opt-in choice for all subscribing users.
Turley committed to ensuring that existing models would remain available with new introductions, assuring users that 4o would not be removed again without advance warning. “If we ever did decide to retire 4o,” he promised, “we’d ensure that users are notified about the timing and process, similar to our current communication practices for the API and enterprise clients.”
Despite the controversies around last week’s update, Turley noted a rise in ChatGPT usage since the rollout of GPT-5. “This creates a unique challenge when you cater to diverse user groups,” he commented. “You can face critiques from engaged power users who rightfully express their concerns, while also addressing the experience of new users encountering advanced capabilities for the first time.”
He stressed the importance of predictability during significant changes, highlighting that OpenAI already provides such predictability for enterprise users. “By applying these principles more broadly, we can enhance user experience,” he stated.
Listeners can hear the full interview with Turley on the upcoming episode of Decoder, airing Thursday, August 14th.
