OpenAI resets Codex usage limits across all plans after a rate-limit spike

• Tibo Sottiaux said OpenAI reset Codex usage limits across all plans after internal dashboards showed more users than expected hitting caps, and he added that the team still did not fully understand the trigger.
• OpenAI's Codex app launch post said Plus, Pro, Business, Enterprise, and Edu plans were already on doubled Codex rate limits, so this reset landed on top of an earlier capacity expansion, as Sottiaux framed it, while growth was still climbing.
• User screenshots from Rhys Sullivan and jxnlco show the reset restoring the 5-hour and weekly meters to 100%, with one screenshot also showing a separate code review bucket.
• A same-day GitHub issue about weekly limit resets shows the flip side: resets can help active users immediately, but they also disrupt anyone treating the weekly window as a predictable budget.
• OpenAI's credits help page says paid credits kick in only after included plan usage is exhausted, which makes these repeated free resets a direct capacity and pricing signal, not just a UX tweak.

You can trace the broader setup in OpenAI's Codex app announcement, which paired the February launch with doubled limits across paid plans. The most revealing artifact is the GitHub complaint about arbitrary weekly resets, because it shows some users were already planning work around rolling weekly budgets before April 1. Then Rhys Sullivan's screenshot and jxnlco's Chinese-language screenshot show the exact counters that got topped back up, including the separate code review meter.

Sottiaux's explanation was unusually direct: more Codex users were suddenly hitting rate limits, OpenAI did not yet fully understand why, and the team chose to reset everyone anyway. In the same post, he said a batch of fraudulent accounts had been banned, which freed some compute.

That timing matters because the official Codex app launch post had already advertised doubled rate limits across Plus, Pro, Business, Enterprise, and Edu. A plan-wide reset a few weeks later suggests the new ceiling still was not matching real demand.

The screenshots make the limit structure more concrete than the announcement did. They show three separate gauges in the product UI:

• a 5-hour usage limit
• a weekly usage limit
• a code review limit

Rhys Sullivan's image shows the session and weekly bars back at 100 percent after the reset. jxnlco's screenshot shows the same pattern in another account, plus the code review bucket restored to full.

The April 1 reset did not come out of nowhere. A widely shared repost claimed Sottiaux had reset Codex usage limits six times during March, and Sottiaux replied a few hours later that Codex growth was surging before the following week's expected bump.

The cleanest external corroboration is a GitHub issue filed the same day. The user said their weekly meter jumped from 60 percent remaining until April 3 to 100 percent until April 8, which wiped out the surplus they had been intentionally banking across the week.

By April 1, users were already talking about a Codex credits promo ending that day, as Kolt Regaskes put it while urging people to burn through credits before the 2x window closed. OpenAI's current flexible credits help page says plan usage is consumed first and paid credits only start drawing after the included limit is hit.

That makes the reset more than a goodwill gesture. OpenAI temporarily gave every plan holder fresh included capacity even though it already sells a paid overage path, and it did so while saying the team still did not know what was pushing so many people into the cap.