Pro plan 5-hour limit is consumed disproportionately compared to weekly limit usage
Description
I am a ChatGPT Pro user and I noticed a serious usage accounting issue with the Codex limits.
My current usage situation:
- Plan: Pro (20x)
- Weekly limit usage: only around 15% used
- 5-hour limit: already fully exhausted (100% used)
Based on the expected limit relationship, the 5-hour limit should normally represent approximately 20% of the weekly limit capacity. Therefore, reaching 100% of the 5-hour limit should correspond roughly to consuming around 20% of the weekly allowance.
However, my account reached the full 5-hour limit while the weekly usage was only at 15%, which suggests the 5-hour usage accounting may be significantly over-counting or not aligned with the weekly quota calculation.
Expected behavior
The 5-hour rolling limit and weekly limit should have a consistent conversion ratio. For Pro 20x users, consuming the entire 5-hour allowance should approximately consume the corresponding portion of the weekly allowance.
Actual behavior
The 5-hour limit was exhausted while the weekly limit showed only ~15% consumption.
Impact
This makes the effective Codex usage capacity much lower than advertised because users can unexpectedly hit the short-term limit even when their weekly allowance is still mostly available.
Could you please investigate whether there is a bug in the Codex usage meter, quota calculation, or the synchronization between the 5-hour and weekly limits?
Thank you.
This issue has 1 comment on GitHub. Read the full discussion on GitHub ↗