On Sunday, OpenAI CEO Sam Altman announced an adjustment to the rate limits for reasoning models, which will impact all users of the San Francisco-based technology firm. While this change will temporarily increase accessibility for various model classes, it is not a permanent measure. The company indicated that it would need to evaluate this capacity trade-off moving forward, hinting that users on the free tier may face some limitations in other features.
Enhanced Reasoning Capabilities for ChatGPT Users
In a message posted on X (formerly Twitter), Altman confirmed that all ChatGPT users would benefit from higher rate limits for the reasoning models. He mentioned that “All model-class limits will shortly be higher than they were before GPT-5.”
The announcement follows an interesting turn regarding the company’s models. With the introduction of GPT-5, OpenAI initially phased out older models. However, due to user dissatisfaction, the company reinstated the GPT-4o model over the weekend. Users had expressed concerns that the new large language models (LLMs) were delivering shorter and less engaging conversational responses compared to their predecessors.
Currently, users on the free tier have access only to the GPT-5 Thinking model, which is integrated into the standard AI setup. Meanwhile, paid subscribers enjoy a distinct Thinking model available for selection in the model picker, in addition to the reasoning capabilities provided by the GPT-4o model, an o-series variant.
Implementing higher rate limits for these reasoning models comes at a significant cost for OpenAI, which has been grappling with operational losses. Altman noted that managing this increase would necessitate some “capacity trade-offs.” He indicated that further details on how the company plans to address these challenges will be forthcoming.
Discussing the rationale for the enhanced rate limits, Altman stated that there has been a noticeable increase in the percentage of users accessing reasoning models daily. He revealed that the proportion of free users utilizing reasoning models rose from one percent to seven percent, while Plus subscribers saw an increase from seven percent to 24 percent.
“I expect use of reasoning to increase over time greatly, so rate limit increases are important,” added Altman.
