OpenAI restores GPT-4o after user backlash over GPT-5 launch
Following an intense user outcry, OpenAI has reinstated its popular GPT-4o model across ChatGPT, marking a swift reversal just days after its controversial removal. On Tuesday, OpenAI CEO Sam Altman announced the model’s return for all paid ChatGPT users, including those with Plus accounts, making it the default option in the platform’s model selection interface. This move directly addresses a significant user revolt that erupted last week when the company, without warning, eliminated all previous AI models upon the launch of GPT-5.
Altman acknowledged that OpenAI had “underestimated how much some of the things that people like in GPT-4o matter to them.” The decision to force users onto the new GPT-5 model on August 7 had triggered one of the most vocal backlashes in ChatGPT’s history. A Reddit thread titled “GPT-5 is horrible” quickly amassed over 2,000 comments within days, vividly illustrating the depth of user dissatisfaction.
Beyond bringing back GPT-4o, OpenAI has implemented several other changes to placate its user base. Rate limits for GPT-5’s “Thinking” mode, which dictates the number of messages a user can send, have been substantially increased from 200 to 3,000 messages per week. Users reaching this new cap can also access additional capacity through “GPT-5 Thinking mini.” Furthermore, the company introduced new routing options – “Auto,” “Fast,” and “Thinking” – providing users greater control over which GPT-5 variant processes their queries.
For ChatGPT Pro subscribers, who pay $200 monthly for enhanced access, Altman confirmed that additional specialized models, including o3, 4.1, and GPT-5 Thinking mini, will become accessible via a “Show additional models” toggle within the ChatGPT web settings. However, he noted that the GPT-4.5 model would remain exclusive to Pro users, citing the high computational costs associated with its operation.
OpenAI also directly addressed widespread complaints regarding GPT-5’s output style. Users had described the new model as “abrupt and sharp,” a stark contrast to GPT-4o’s more conversational and nuanced tone. Some users even expressed profound grief over losing what they considered their “only friend,” highlighting an unexpected emotional bond with the older AI. Altman stated that the company is “working on an update to GPT-5’s personality which should feel warmer than the current personality but not as annoying (to most users) as GPT-4o.” He also indicated a future direction toward “more per-user customization of model personality.”
The initial GPT-5 launch was marred by a series of issues extending beyond the model removal. An automatic routing system, designed to intelligently select appropriate model variants, malfunctioned on launch day, consistently defaulting to less capable versions. OpenAI also faced criticism for including misleading performance graphs in its launch presentation, an error Altman later admitted was a “mega chart screwup.”
While GPT-4o has now returned, OpenAI continues to refine GPT-5, acknowledging that further adjustments to rate limits may be necessary “depending on usage” and that personality updates are still under development. For now, paid ChatGPT users can once again choose their preferred model, representing a significant compromise between OpenAI’s strategic push for newer AI models and the vocal demands of its user community for choice and familiarity.