codex - 💡(How to fix) Fix Codex Desktop five-hour usage-limit percentage sometimes increases during active window

Official PRs (…)
ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

Helpful · Quick feedback

Loading…
RAW_BUFFERClick to expand / collapse

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

Codex Desktop, current build observed on 2026-06-04 / 2026-06-05.

Exact app version was not captured yet.

What subscription do you have?

Unknown / not confirmed. The issue is visible in the Codex Desktop usage-limit UI.

What platform is your computer?

Windows desktop app.

What issue are you seeing?

The five-hour usage-limit percentage sometimes increases during an active usage window.

This is separate from a freshly reset window starting at 99% instead of 100% (#26370). That issue is about the starting value after reset. This issue is about the displayed remaining percentage moving upward between observations within an active five-hour window.

Observed examples:

  • The five-hour meter moved from roughly 47% remaining to about 50% or 51% remaining.
  • The five-hour meter later moved from 10% remaining to 11% remaining.

Both observations were on the five-hour window.

I did not capture the exact raw account-rate-limit payloads for these transitions, so this is not yet a minimal reproduction. However, the user-facing behavior is confusing: a remaining-capacity meter should not increase while work is being performed unless the five-hour window has actually reset.

What steps can reproduce the bug?

Exact reproduction steps are not known yet.

Observed path:

  1. Use Codex Desktop during an active five-hour usage window.
  2. Watch the usage-limit UI across refreshes.
  3. Compare the displayed five-hour remaining percentage before and after a refresh / subsequent work.

Observed result:

  • The five-hour remaining percentage can increase, for example from about 47% to 50% / 51%, or from 10% to 11%.

Expected result:

  • During one active five-hour window, the remaining percentage should either stay the same or decrease as usage is consumed.
  • It should increase only when the five-hour window has actually reset or when the UI explicitly indicates that it is showing a different bucket/window.

What is the expected behavior?

The Codex Desktop five-hour usage-limit meter should be monotonic within a single active window from the user's perspective.

If the displayed value is "remaining capacity", it should not move upward between refreshes unless a reset occurred.

If the backend can legitimately return a higher remaining percentage due to delayed accounting, bucket reconciliation, stale snapshots, or window changes, the UI should make that state clear rather than presenting it as ordinary remaining capacity for the same five-hour window.

Additional information

Related but distinct issue:

  • #26370 reports the five-hour usage-limit meter starting at 99% instead of 100% after reset.

Current open-source context, if Desktop is using the same or similar account-rate-limit snapshot contract:

  • The CLI/TUI status path maps backend used_percent into displayed remaining capacity as 100 - used_percent.
  • The rate-limit poller refreshes account snapshots periodically.
  • A visible increase in the displayed "left" percentage could indicate that backend used_percent decreased between snapshots, the UI switched between stale/fresh or otherwise inconsistent snapshots, or the display layer rounded/clamped adjacent values in a way that made remaining capacity appear to increase.

The core concern here is the user-facing invariant: within a single five-hour window, a remaining usage percentage should not visibly go up without an actual reset or an explicit explanation.

Vote matrix · Quick signals

Works
Did the solution work? Tap to confirm.
Easy Fix
Was it a quick fix?
Time Saver
Did it save you time?
Blocking
Was it severely blocking?
Common Issue
Are others likely hitting this too?
Flaky / Intermittent
Is it intermittent?
Verified / Reproducible
Can you reproduce it reliably?
Loading…

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Back to top recommendations

TRENDING