Weekly limits dropping too quickly

Open 💬 3 comments Opened Apr 29, 2026 by Bobushka
💡 Likely answer: A maintainer (github-actions[bot], contributor) responded on this thread — see the highlighted reply below.

What version of the Codex App are you using (From “About Codex” dialog)?

Version 26.422.71525 (2210)

What subscription do you have?

PRO

What platform is your computer?

Darwin 25.4.0 arm64 arm

What issue are you seeing?

The weekly limit dropped by 4% in the new chat on an empty project after the first prompt "Hi" and 8k tokens used. and this is not the only case: this behavior started yesterday and my weekly limit dropped in 2 hours of normal work from 100% to 22%.

What steps can reproduce the bug?

019dd844-21c2-79c2-9567-8677bc103bee
258k
5.5 high

What is the expected behavior?

Typically, this amount of work takes up 2-3% of the weekly limit, even on a new model.

Additional information

_No response_

View original on GitHub ↗

3 Comments

github-actions[bot] contributor · 4 months ago

Potential duplicates detected. Please review them and close your issue if it is a duplicate.

  • #20153
  • #19996
  • #19989
  • #18699
  • #18696

Powered by Codex Action

Bobushka · 4 months ago

It's funny that even when I'm not doing any work at all, just with the app open, the weekly limit decreases by 1% every 6 hours. Is this normal?

Bobushka · 4 months ago

This is not a UX issue or misunderstanding.

This is a billing / metering failure affecting a paid PRO subscription.

Key facts:

  • Empty project (no repo, no files, no agents)
  • First prompt "Hi" → -4% weekly limit (~8k tokens)
  • Idle drain: -1% every ~6 hours with no interaction
  • Started suddenly (date: 2026/04/28)

This makes the product unusable:

  • Usage decreases without user actions
  • Costs are not predictable or controllable
  • Remaining quota is being consumed passively

This should be treated as:

  • Billing bug
  • Severity: High
  • Impact: Paid users losing quota without usage

Please confirm:
1) Is this expected behavior?
2) If not — is this a known incident?
3) Is usage being incorrectly metered in idle sessions?

Request:

  • Escalation to engineering
  • Status update or incident acknowledgment