Codex Resets
This site diligently tracks OpenAI's Codex and ChatGPT Work usage limit resets, offering a fascinating peek behind the curtain of rapid AI development. It compiles public announcements from an OpenAI team member, revealing frequent resets driven by exponential user growth, system incidents, and new feature rollouts. The tracker provides rare, data-driven insight into the operational complexities of scaling bleeding-edge AI services.
The Lowdown
The "Codex Resets" website serves as a dedicated tracker for usage limit resets pertaining to OpenAI's Codex and ChatGPT Work platforms. It aggregates announcements, primarily from an OpenAI employee on X, detailing each instance when these limits are replenished for users. The site provides a real-time, historical log of these events, painting a picture of the dynamic environment surrounding highly demanded AI services.
- The most recent reset occurred just 20 hours prior to the snapshot (July 18, 2026, 3:28 AM).
- A total of 35 resets have been recorded, with an average interval of 8.9 days between them.
- The longest period without a reset stretched to 67.7 days, indicating less frequent disruptions or changes in earlier periods.
- Resets are frequently tied to significant user milestones (e.g., reaching 7M, 8M, 9M active users), system incidents, or outages requiring mitigation.
- Other common reasons include the launch of new features, plugins, or pricing plans (such as GPT-5.6 Sol or a new $100 plan).
- Occasionally, resets are promotional, celebratory, or intended to compensate for identified issues like usage being drained faster than expected.
- The announcements often adopt a casual, transparent tone, sometimes mentioning the internal challenges of scaling infrastructure and crediting teams for their rapid response.
This tracker provides a valuable, albeit unofficial, public record of the operational tempo and challenges faced by OpenAI in managing its popular AI development and work platforms.