OpenAI's Codex team is digging into sudden token-usage spikes that have been burning through weekly limits far faster than users expect on unchanged projects. Engineering lead Thibault Sottiaux confirmed the group held an active weekend review of logs while applying immediate resets to restore access. A team member lightened the mood by jokingly taking the blame for the extra consumption.
Codex team is 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. Taking it very seriously and won't rest until we get to the bottom of it.
i think openai has gotten a bit too loose and funny acting. i think its time to get a little serious
Codex users saw their caps restored after the unexplained drain, giving everyone breathing room while the investigation continues. The move addresses both the current spike and any banked resets that had piled up.
No confirmed cause has surfaced yet for the doubled or higher consumption rates, leaving open whether model updates, agent behavior, or logging systems are at play. The team has not indicated how long the resets will last or if similar spikes could recur.
Positive users thank Tibo for promptly resetting Codex usage limits and investigating drains on a Sunday, while negative users complain about wrongful account bans and blame on customers.
No Digg Deeper questions have been answered for this story yet.
As we are still investigating, I have reset everyone's Codex usage limits. This is a hard reset given some users had stacked up to three banked resets already that they can apply on their own schedule.
Funnily enough, this week at OpenAI is called the RESET week, which is meant for folks to relax a bit. However it will be a different kind of RESET week. Enjoy.
Codex team is 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. Taking it very seriously and won't rest until we get to the bottom of it.
If you happened to have used your previous reset in the few hours before and didn't go through your usage, do not worry, you will get more manual resets after we conclude the investigation.
As we are still investigating, I have reset everyone's Codex usage limits. This is a hard reset given some users had stacked up to three banked resets already that they can apply on their own schedule.
Funnily enough, this week at OpenAI is called the RESET week, which is meant for folks to relax a bit. However it will be a different kind of RESET week. Enjoy.
@solipsnitsyn Auto review is using GPT-5.4. We do not route. Do not spread misinformation, thank you.
@thsottiaux thank you tibo. theres a reason ive always been and will always be team openai
@thsottiaux while totally ignoring that you've been routing us to 5.4 for months without saying anything, yeah, we know, I'm about to file a chargeback over it
@thsottiaux Thanks also please look to 5.5 xhigh, because it is actually fucking dumb.
@thsottiaux Do you guys ever rest?
@thsottiaux I used my reset 1 hour ago man … 😪
@thsottiaux FYI I think it’s Chronicle. I had the same large drain on usage, but Codex troubleshoot that Chronicle was the driver. Once I paused that, usage drain was normal
@thsottiaux it's me consuming all the tokens tibo im sorry i can spare a few but they are too good
As we are still investigating, I have reset everyone's Codex usage limits. This is a hard reset given some users had stacked up to three banked resets already that they can apply on their own schedule.
Funnily enough, this week at OpenAI is called the RESET week, which is meant for folks to relax a bit. However it will be a different kind of RESET week. Enjoy.
@thsottiaux Inside OpenAI right now:
@LLMJunky @thsottiaux This week was supposed to be a chill week
@LLMJunky @thsottiaux we rest wen ded
It’s odd that Claude Code (and Claude models in general) consume far more tokens and cost more than OpenAI models. I can run Claude Code Opus 4.8 on the Fast tier all day without hitting my 5‑hour or weekly limits, yet when I use Codex with GPT‑5.5, supposedly a more efficient model that also runs on Fast, it exhausts my 5‑hour limit in just a few minutes.
Both services are on the Fast mode, but Claude Code never reaches the limit, even if I try to push it. OpenAI’s past promotions (e.g., 2× extra usage) are no longer relevant. Right now the situation feels unusable: the numbers, usage, and limits don’t line up. I have a 20× plan for both Claude and OpenAI, but Claude usage feels like it’s actually 20× higher, while OpenAI usage burns through the allowance much faster.
If you can investigate what’s happening, I’d appreciate it. If there’s nothing wrong, please let me know. It just seems like the token consumption is disproportionately high.
@thsottiaux i just used my banked reset an hour ago...
@thsottiaux I'm having some issues with my $200 pro subscription. I've reset it twice in just one week, and the current one is almost about to run out. My usage has increased significantly, like four times the normal amount.
@ShashankDe5535 You'll have to wait a bit to enjoy that
@thsottiaux I get one or two prompts now, total, before 5 hour is gone. My weekly is absorbed so fast. It's been this way for too long and I know I'm not the highest paying customer at the Plus level, but recognize that someone at this level is not eating by spending that $20 a month.
@thsottiaux I just used my reset few minutes ago 😭😭
@thsottiaux And I even went to 5.4 medium and same results.