OpenAI CEO Sam Altman has announced a temporary increase in the rate limit of the company’s reasoning models for all ChatGPT users, in response to a surge in demand for these capabilities.
This change affects all model classes and comes after user backlash led to the reinstatement of the GPT-4o model over the weekend. According to Altman, “All model-class limits will shortly be higher than they were before GPT-5,” a statement that holds significance given OpenAI’s previous retirement of older models with the release of GPT-5. The reinstatement of GPT-4o for Plus subscribers followed user complaints about the brevity and perceived lack of “warmth” in conversational responses from the latest large language models (LLMs).
Currently, free-tier users primarily access the GPT-5 Thinking model, integrated into the standard GPT-5 AI model, which unifies the GPT- and o-series models. In contrast, paid subscribers have access to a distinct Thinking model, selectable via a model picker, and can also utilize GPT-4o, possessing reasoning capabilities. The decision to increase rate limits for these models will incur significant operational costs for OpenAI, a company reportedly already running at a loss.
Altman noted that managing the increased rate limit will necessitate “capacity trade-offs,” with the company planning to disclose how it will address these challenges in the near future. He hinted that free-tier users might experience a reduction in other features as a consequence.
The primary driver behind this adjustment is the growing adoption of reasoning models. Altman shared statistics highlighting a substantial increase in daily users accessing these capabilities, with the percentage of free users utilizing reasoning models surging from one percent to seven percent, and from seven percent to 24 percent for Plus subscribers.
Altman expressed his expectation for continued growth in reasoning model usage, stating, “I expect use of reasoning to increase over time greatly, so rate limit increases are important.” OpenAI’s future strategies regarding capacity allocation will be crucial in balancing user demand with operational sustainability.
