Skip to content

[Usage][Luna] Weekly Codex limit dropped ~40% in one day despite light usage #39818

Open

Description

@nightcat01

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

Latest Windows Codex App update available on 2026-08-21. Exact build number was not captured before filing.

What subscription do you have?

ChatGPT plan with Codex access.

What platform is your computer?

Windows desktop Codex app.

What issue are you seeing?

After the recent Codex update, my weekly Codex usage dropped from roughly 100% remaining to about 60% remaining in a single day, despite doing comparatively little work and using GPT-5.6 Luna.

The amount of work performed was less than half of a previous period in which I eventually consumed the full weekly allowance. The current session therefore appears to be consuming quota far faster than my established baseline.

This is particularly unexpected because the work was performed with Luna, which is intended to have substantially lower usage/credit cost than Sol.

What steps can reproduce the bug?

  1. Open Codex on Windows after the latest available update.
  2. Select GPT-5.6 Luna.
  3. Perform normal development work on an existing project.
  4. Avoid intentionally heavy usage such as large-scale repository-wide refactors or prolonged high-cost model runs.
  5. Check weekly usage before and after the work session.
  6. Observe a much larger drop than expected relative to prior usage patterns.

What is the expected behavior?

Luna usage should consume the weekly Codex allowance materially more slowly than Sol for comparable work, and normal/light development activity should not reduce the weekly allowance by approximately 40 percentage points in one day when the workload is substantially below a previous full-week workload.

Actual behavior

  • Approximately 40% of the weekly allowance was consumed within about one day.
  • Remaining allowance is approximately 60%.
  • The workload was less than half of a prior period that eventually consumed the entire allowance.
  • GPT-5.6 Luna was used for the work.
  • The perceived consumption rate is dramatically higher than the user's prior baseline.

Additional information

This may be related to existing usage-accounting reports, but I am filing a separate report because the magnitude and timing are easy to distinguish from my prior baseline.

Potentially related:

I cannot confirm whether the cause is model weighting, background activity, or server-side usage accounting. The main reproducible observation is the sharp change in weekly-limit consumption immediately after the recent update while using Luna for a relatively light amount of work.

If useful, please expose per-session/per-model quota attribution so users can distinguish foreground Luna requests from background/sub-agent/tool activity and verify what consumed the weekly allowance.

Activity

  1. added
    bugSomething isn't working
    appIssues related to the Codex desktop app
    windows-osIssues related to Codex on Windows systems
    rate-limitsIssues related to rate limits, quotas, and token usage reporting
    on Aug 20, 2026
  2. aqua5230 commented on Aug 24, 2026

    @aqua5230

    Similar reports in #36848. If it helps document the pattern: usage is a free tray app that tracks your weekly quota history locally — might make it easier to show OpenAI exactly when the drop hit.

  3. jin44194649 commented on Sep 30, 2026

    @jin44194649

    Related current data point from Sep 29, 2026 (America/Chicago): on a $200 ChatGPT Pro account, the weekly meter reset/upgrade time was 18:09. The main ChatGPT Work session lasted about three hours (to ~21:09) and was used for eight chapters of writing/review. By 22:32 the weekly meter showed 89% remaining. I am not claiming the same root cause as this issue, but the effective consumption was unexpectedly high and the UI did not provide enough task-level attribution to reconcile the drop. I added the fuller timeline in #40527.

    The overlap is unexpectedly high weekly usage relative to the user-visible workload. This case was Work rather than Luna/Codex, so it may be a different root cause, but it would be useful to correlate plan-level weighting/accounting changes across surfaces.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    appIssues related to the Codex desktop appbugSomething isn't workingrate-limitsIssues related to rate limits, quotas, and token usage reportingwindows-osIssues related to Codex on Windows systems

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions