A Community Tracker Counts OpenAI's 35 Codex Usage-Limit Resets

Codex Resets is a fan-built dashboard that logs every time OpenAI resets usage limits for its Codex coding tool and ChatGPT Work tier, drawing from the frequent, ad-hoc announcements posted on X by OpenAI’s Thomas Sottiaux. The site tallies 35 resets to date, arriving on average every 8.9 days, with a heat-map calendar and a running archive of the announcement text behind each one.

The logged messages reveal the pattern behind the giveaways. Some resets are pure marketing—celebrating user-count milestones from 500K up past 9M, a new $100 plan, a one-year anniversary, or the launch of plugins and new GPT-5.x models. Others are damage control tied to reliability incidents: requests wrongly blocked as “high cyber risk,” a caching optimization that drained limits faster than expected, promotional 2x limits that failed to apply to a slice of users, and repeated investigations into complaints that Codex was burning through quotas too quickly. OpenAI has also rolled out “banked resets” that users can trigger themselves from web, mobile, or desktop.

The casual, meme-laden tone of the posts—one admits resets “cost money” but “the vibes are good”—captures a company scaling infrastructure faster than it can keep stable, and using free usage as both apology and hype engine. For technical readers, the tracker is a useful proxy for how often OpenAI’s coding platform hits capacity or billing problems, and how aggressively the company leans on generosity to smooth over the disruption.