Codex weekly limits just reset early - Plus and Team accounts this time
Weekly limits for GPT-5.3 and 5.4 unexpectedly refreshed, giving users a fresh start.
OpenAI has executed another unexpected early reset of usage quotas for its Codex API, this time affecting both ChatGPT Plus and Team subscription tiers. The reset cleared all three primary quota metrics: the 5-hour limit dropped to 0.0%, the weekly all-model usage reset to 0.0%, and review requests were restored to 100%. This follows a similar incident last week that only impacted Plus users, suggesting either a systematic backend adjustment or an intentional policy change by OpenAI.
While the company hasn't officially commented on whether this is intentional or a technical quirk, the practical effect is immediate for developers. Those who were approaching their weekly limits on models like GPT-5.3 and GPT-5.4 now have a fresh allocation, allowing continued development and testing. The community-developed tracking tool OnWatch (https://github.com/onllm-dev/onwatch) helps users monitor these quota fluctuations in real-time, providing visibility into OpenAI's often-opaque usage management system.
- Early quota reset affected both ChatGPT Plus ($20/month) and Team ($25/user/month) tiers
- Cleared 5-hour limit and weekly all-model usage to 0.0%, restored review requests to 100%
- Follows previous reset that only impacted Plus users, suggesting broader system changes
Why It Matters
Developers get unexpected capacity for GPT-5.3/5.4 testing, but quota instability creates planning challenges.