OpenAI Codex says that the abnormal weekly limit consumption affected too few users to justify a global reset. If you’ve experienced unusually fast use of your weekly limit, please report it on the dedicated issue page.

OpenAI is responding to user complaints about its Codex API, where some developers report their weekly usage limits are being consumed abnormally fast due to a suspected bug. In a communication, OpenAI stated the issue affected "too few users to justify a global reset," opting instead for a case-by-case resolution. This has sparked frustration in the developer community, with advocates arguing the problem is more widespread but underreported, as many users may not know how to formally notify OpenAI. The company is directing impacted users to a specific GitHub issue page (#13568) to detail their experience and provide account information for manual limit resets.

The technical specifics of the bug remain unclear, but it highlights the challenges of managing consumption-based API services like Codex, which powers GitHub Copilot. For developers and companies, unexpected limit exhaustion can disrupt workflows and incur unplanned costs if they need to upgrade tiers prematurely. The community-led effort to consolidate reports on GitHub aims to provide OpenAI with clearer data on the issue's scope, pressuring for a more systemic fix rather than individual patches. This incident underscores the growing need for transparent communication and robust monitoring tools as AI APIs become critical infrastructure for software development.