OpenAI warroom after Codex limits drain too fast

OpenAI warroom – OpenAI opened a “warroom” to investigate why some users say Codex usage limits are depleting faster than expected. The company said the issue is tied to its “abuse and fraud prevention systems” incorrectly rate limiting certain accounts, and that it reset user

For the second day in a row, Adam watched his Codex allowance vanish like it had been spilt—each of the last two days burning through an entire week’s usage on the $200 plan. He said he had to use a reset for the first time, even though similar tasks used to last longer.

Across the weekend, that kind of experience kept surfacing on X. OpenAI then stepped in with a move that sounded more like an emergency room than a routine product update: it set up a “warroom” to investigate a flurry of user reports that its coding agent was hitting usage limits faster than usual.

On its status page. OpenAI said some users’ Codex limits are “depleting faster than expected.” It traced the problem to its “abuse and fraud prevention systems” incorrectly rate limiting certain accounts. The stakes are straightforward for working engineers: Codex usage limits track how much compute a coding task uses. and the limits are shown on a user’s dashboard as a percentage.

More intense tasks burn through credits faster, and the limit varies by subscription tier. But the reports going around over the weekend described a mismatch between expected behavior and what users were seeing for comparable work.

Thibault Sottiaux. engineering lead for OpenAI’s Codex. said on Sunday that OpenAI implemented an across-the-board reset of user caps while it investigated. In an X post dated Sunday. he said the Codex team was “in a warroom on a Sunday combing through logs and checking whether there is anything that could lead to increased usage drains for some users.” He added. “Taking it very seriously and won’t rest until we get to the bottom of it.”.

He also explained the reset mechanics. “As we are still investigating, I have reset everyone’s Codex usage limits,” he wrote on Sunday. “This is a hard reset given some users had stacked up to three banked resets already that they can apply on their own schedule.” In the same post. he referenced that the current week at OpenAI is called the RESET week.

One user’s account captured the frustration in plain terms: Adam wrote that he had to work hard the entire week to burn through the 7 days’ usage, but that the last two days burned the entire week’s usage in one day each. He said he then had to use the reset for the 1st time.

OpenAI’s status page said the impact appears limited and that it was continuing to monitor the situation.

Zooming out. this incident lands in a broader shift across AI tooling: companies have been reducing usage limits as surging popularity strains compute resources. In March. Anthropic adjusted Claude usage caps during peak hours. and some engineers reportedly reshaped their workdays while they waited for limits to reset.

More generally. AI providers have moved away from all-you-can-eat access. and some companies have started limiting employee AI use to keep costs down. OpenAI’s Codex limit issue follows an outage earlier this month. Anthropic has had similar problems. and a Claude outage in March had underscored how reliant software engineers have become on these tools—some bemoaned the need to write code by hand again.

The sequence here is what users are likely to remember: a weekend of complaints that limits were falling unusually fast. followed by OpenAI’s “warroom. ” an explicit explanation tied to its abuse and fraud prevention systems. and then a hard reset of Codex usage limits across the board while the logs are combed for answers.

OpenAI Codex usage limits warroom abuse and fraud prevention rate limiting software engineers AI compute subscription tiers Anthropic Claude outage