- Published on
Codex Resets
- Authors
- Name
- DeepHorus
OpenAI has been issuing frequent usage limit resets for its Codex and ChatGPT Work platforms, as tracked by a public log maintained by the company's team. Since the tracking began, there have been 35 total resets, with an average interval of 8.9 days between them. The longest gap between resets was 67.7 days, indicating periods of stability punctuated by sudden bursts of activity. These resets are often announced unexpectedly on social media, typically by OpenAI employee @thsottiaux on X.
Codex is an AI coding assistant, while ChatGPT Work is a premium productivity tier. The resets are frequently tied to user milestones, such as reaching 7 million, 8 million, or 9 million active users. In one announcement, the team noted hitting 9 million active users "way earlier today" but got distracted by infrastructure scaling issues. Another reset celebrated 7 million active users with a "banked reset" feature that allows users to store additional resets for later use.
The resets have become a notable pattern: some are hard resets that replenish weekly usage limits, while others include extra "banked" resets that users can apply on their own schedule. One post mentioned that some users had stacked up to three banked resets. The team also acknowledged technical issues, such as reports that Codex usage was being consumed faster than expected, though no single central issue was found.
These frequent resets suggest that demand is outpacing OpenAI's ability to maintain consistent usage caps. The company's infrastructure is under strain, as indicated by mentions of "keeping the infra up as we scale faster than ever" and "three separate small incidents that affected Codex reliability" in one 24-hour period. The resets serve as a temporary fix to ensure users can continue working without interruption.
Community reactions have been mixed. Some users appreciate the extra capacity, while others note the unpredictability of the resets. A reset announcement on a Friday encouraged users to "explore the boundaries of GPT-5.6 Sol," showing OpenAI's push for ambitious experimentation. However, the reliability concerns have led to investigations and mitigations being deployed.
Looking ahead, OpenAI is actively working on system improvements. One post revealed that the company is "iterating at lightspeed" and hiring, such as new model researcher @_rajanagarwal to push on coding capabilities. The "RESET week" was even internally themed, indicating the resets are a known operational pattern. As user numbers climb, sustainable usage management solutions will be critical for long-term platform stability.
