Hi everyone,itle: Possible abnorm
I am a ChatGPT Pro / Pro Coding Plan user, and I rely on Codex every day for my work. I want to report a possible abnormal usage-metering issue with the Pro Coding Plan weekly allowance.
This is not a dashboard loading issue, browser issue, cache issue, VPN issue, or extension issue. The Usage Dashboard loads normally. The problem is that the weekly usage allowance appears to be consumed at an abnormal rate compared with my own historical usage pattern.
Here is the comparison:
Before the recent weekly refresh, my previous 7-day usage cycle looked normal. I used Codex heavily every day, often with hundreds of rounds per day, and across the full 7-day period the total usage reached several thousand rounds before the Pro weekly allowance was nearly consumed. That matched my normal work pattern as a Pro user.
However, after the weekly allowance refreshed yesterday, the behavior changed suddenly. I only used Codex lightly for about one to two days, and the recorded rounds were much lower than before. In one example from the Usage Dashboard, the recorded usage for one day was only around 121 total rounds across models: 49 for gpt-5.5, 47 for gpt-5.4, and 25 for gpt-5.4-mini.
Despite this much smaller amount of recorded usage, the weekly allowance was already about 70% consumed by the second day of the cycle. In other words, the current usage volume is not even close to one of my previous heavy-use days, but it has already consumed most of the weekly Pro allowance.
I understand that Codex usage may not be calculated only by round count. Model type, token volume, task size, reasoning, fast mode, instances, and other factors may affect usage. However, the difference here is too large compared with my own previous usage on the same Pro account and similar workflow. My usage environment and configuration did not change significantly.
The key issue is the abnormal consumption rate:
Previous cycle: several thousand rounds over seven days were needed to nearly consume the Pro weekly allowance.
Current cycle: after only one to two days and much lower recorded usage, around 70% of the weekly allowance was already consumed.
This suggests a possible backend usage-metering, rate-limit calculation, entitlement, or statistics issue affecting the Pro Coding Plan.
I have already contacted support, but the response focused on browser/cache/VPN/dashboard troubleshooting. That does not address the issue. The dashboard is visible; the concern is whether the server-side metering is counting usage correctly.
Could someone from OpenAI or the Codex team please help review whether there is a known issue with Pro Coding Plan weekly usage metering, or advise how to get this escalated for a manual backend review?
I can provide screen recordings, timestamps, screenshots of the Usage Dashboard, and the relevant support ticket privately if needed.
Thank you.
