Tibo Reset Codex Usage Three Times This Week, but OpenAI's Limit Rules Haven't Changed

OpenAI's Tibo Sottiaux reset Codex usage limits three times this week, on October 13, 15 and 18, Beijing time. For anyone on a ChatGPT plan, that raises two questions: is it worth holding heavy work for the next one, and has the way limits are counted changed? Going by Tibo's posts and OpenAI's documentation, nobody has promised another reset, and the rules haven't changed.

Three resets, little explanation

Each began as a numbered entry in Tibo's daily update thread on X. On Tuesday, October 13, he posted "Day 8/ Reset" at 09:03, added "Global reset landing by EOD PST time" seconds later, and wrote "It's done" at 13:06. On Thursday, October 15, "Day 10/ Reset" went up at 12:55 and "Done and dusted" at 14:28. On Sunday, October 18, he posted "Day 13/ Le limit reset" at 00:20 and "Done" at 05:12. That count doesn't include an earlier reset on Sunday, October 11, at the end of the previous week.

Only the first of those calls itself a global reset. None of the posts says whether the weekly allowance, the five-hour window or both were refreshed. OpenAI's help center says an automatic reset covers "the eligible usage limits specified in the announcement," and these announcements didn't specify.

Tibo hasn't explained the pace either. At 02:02 on October 18, X user Ananya Singh asked him, "Why is Codex giving reset every day? 😭" At 06:23 he replied, "Pot committed." In poker, that describes a player who has already put so much into the pot that folding no longer makes sense. He didn't elaborate.

Should you wait for the next one?

OpenAI's help center describes one-time resets as something OpenAI "may occasionally provide," says eligibility and the limits affected vary by offer, and states that "future resets are not guaranteed." Tibo hasn't announced another one. OpenAI hasn't published a schedule for them, and the three this week came two and three days apart.

The date that is fixed is your own. OpenAI says banked and purchased resets move your weekly reset date; its help pages don't say the same about automatic resets. The reset time under Settings and then Usage, or via /status in the Codex CLI, is the one that applies to you.

The rules underneath haven't moved

A reset refills the allowance without changing it. OpenAI's GPT-6 Astra usage guide puts it this way: "A reset restores your allowance. It does not permanently increase your plan's limits or change how much allowance a task uses."

The limits still work as OpenAI described them before this week. Depending on the plan, Codex usage can be capped over a five-hour window and a weekly window, and you need room in both to keep going. The five-hour window isn't tied to the clock: a new one starts with your first Work or Codex message after the previous window ends. Pro plans currently have no five-hour limit but still have an included allowance. OpenAI publishes no weekly figure for any plan, and it says the allowance isn't a fixed number of messages, because the model, context, reasoning, tool use, retrieval and caching all change what a task costs. Its estimate for Plus and Business Standard still runs from 5 to 45 local messages per five hours on GPT-6 Astra to 350 to 3,000 on GPT-6 Luna.

Those figures match what we found when we checked OpenAI's Codex pricing page on October 9, and the Codex changelog entries since then list no change to usage limits. The full history of Tibo's resets is on our Codex Reset Tracker page.