David Bridges
Following a tumultuous 72 hours marked by a significant backlash from its most dedicated users, OpenAI is executing a substantial policy reversal. The company, under the leadership of CEO Sam Altman, took to X (formerly known as Twitter) on Sunday to announce that it is reinstating its cherished older AI models, including GPT-4o, while also significantly boosting usage limits for paying subscribers. This move serves as a clear peace offering to a deeply dissatisfied customer base.
This dramatic decision comes just days after the problematic launch of GPT-5, the company’s latest and most advanced model. What should have been a celebrated achievement instead ignited widespread discontent among users.
the percentage of users using reasoning models each day is significantly increasing; for example, for free users we went from <1% to 7%, and for plus users from 7% to 24%.
i expect use of reasoning to greatly increase over time, so rate limit increases are important.
— Sam Altman (@sama) August 10, 2025
Understanding the Fallout from a Disastrous Model Launch
On August 7, OpenAI introduced GPT-5, branding it as a “unified system” designed to automatically direct user queries to the most suitable model for the task at hand. This shift resulted in the removal of the menu that allowed users to select between older, reliable models like GPT-4o, which had been launched in March 2023. The abrupt change left many users feeling abandoned, especially those who had invested in subscriptions such as ChatGPT Plus at $20 per month. These users had developed their workflows around a diverse set of models, each tailored for specific tasks, such as creativity, logical reasoning, or in-depth research. The forced transition to a single model disrupted their established processes and stripped them of the ability to cross-reference responses, leading to an immediate and vigorous backlash that included subscription cancellations and a flurry of online petitions.
Sam Altman’s Swift Response to User Concerns
As the unrest escalated over the weekend, Sam Altman took to X to perform damage control, openly acknowledging that the company had miscalculated the situation and outlining a series of adjustments aimed at restoring user trust. One of the most significant announcements came in response to a user query that echoed the sentiment of countless others: “And bringing back 4o?”
“It’s back! Go to settings and pick ‘show legacy models’,” Altman enthusiastically affirmed, confirming the return of the long-adored model. Although GPT-5 remains the default option, users now have the choice to revert to older versions that better suit their needs.
it’s back! go to settings and pick “show legacy models”
— Sam Altman (@sama) August 10, 2025
In a lighthearted exchange, a user inquired whether using the legacy models would be considered “marked as weird.” Altman humorously replied, “Not marked as weird in either case!” To further pacify subscribers, Altman announced a substantial increase in usage limits for the advanced features of GPT-5. “Today we are significantly increasing rate limits for reasoning for ChatGPT Plus users, and all model-class limits will shortly be higher than they were before GPT-5,” he stated.
When pressed for details about the new limits, Altman disclosed the new cap: “Trying 3000 per week now!” This remarkable increase for a $20 per month subscription serves as a clear incentive for irate users to remain loyal to the platform.
trying 3000 per week now!
— Sam Altman (@sama) August 10, 2025
Commitment to Enhanced Transparency and User Engagement
In a final gesture of goodwill, Altman committed to improving transparency within the platform. He announced an imminent user interface change that will allow users to see which model is currently processing their queries. Additionally, he promised a comprehensive blog post later in the week that would elucidate the company’s “thinking on how we are going to make capacity tradeoffs.”
Altman also shared compelling data illustrating the growing popularity of the new reasoning models, revealing that daily usage among ChatGPT Plus users has surged from 7% to 24%. This statistic subtly legitimizes the company’s initial emphasis on the new technology, despite the tumultuous rollout. The term “reasoning” refers to the AI’s capability to methodically “think” through complex problems step-by-step.
Ultimately, OpenAI appears to have heeded the feedback from its users loud and clear. This incident showcased a rare and powerful demonstration of user revolt, compelling one of the most influential companies in the tech industry to listen and adjust its strategies. The compromise—keeping GPT-5 as the default while reinstating user choice—stands as a direct acknowledgment of the strength of a dedicated community that refuses to be overlooked.
Daniel Mercer
Brandon Fletcher
