Codex app usage limit depleted abnormally: 5-hour allowance consumed in about 1 hour

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

26.623.31921

What subscription do you have?

ChatGPT Pro subscription with 20x usage allowance.

What platform is your computer?

Darwin 25.5.0 arm64 arm

What issue are you seeing?

I am experiencing an abnormal usage-limit depletion issue in the Codex app.

I have a ChatGPT Pro subscription with 20x usage allowance. During my recent usage, the Codex app consumed what appears to be approximately a 5-hour allowance within about 1 hour. This is significantly different from my previous usage pattern. In similar sessions with a comparable workload and intensity, the same type of usage usually lasted around 3 to 4 hours before reaching the limit.

Actual behavior:

• The usage allowance was depleted in about 1 hour.
• This appears to correspond to roughly 5 hours of expected allowance being consumed in a much shorter period.
• The depletion rate was much faster than my previous similar Codex usage sessions.
• I did not see a clear explanation, warning, or detailed breakdown showing why the allowance was consumed so quickly.

Expected behavior:

• Usage should be deducted consistently and transparently.
• Similar workloads should not suddenly consume the allowance several times faster without a clear explanation.
• If there are different deduction rates based on model, retries, failed requests, background actions, tool calls, or other factors, the app should clearly show this to the user.
• Failed requests, retries, refreshes, or background operations should not cause duplicate or excessive usage deductions.

Why I believe this may be a bug:

• Previous similar usage usually lasted around 3 to 4 hours.
• This time, the allowance was depleted in about 1 hour.
• I did not intentionally change my usage pattern enough to explain such a large difference.
• There was no clear usage breakdown explaining the accelerated depletion.

Please investigate whether there may be:

• duplicate usage deductions;
• failed requests being counted as successful usage;
• automatic retries or background actions consuming allowance unexpectedly;
• incorrect accounting of model/tool usage;
• delayed or inaccurate limit display;
• a regression in the Codex app’s usage tracking;
• missing user-facing disclosure about how usage is calculated.

Time of issue:
[paste date and approximate time here]

Codex App version:
[paste version here]

Subscription:
ChatGPT Pro with 20x usage allowance

Platform:
[paste platform output here]

I understand that account-specific billing or quota restoration may need to be handled by OpenAI Support separately. However, I am filing this issue here because the Codex app appears to have consumed usage allowance at an abnormal rate compared with previous similar sessions, and I would like the engineering team to check whether there is an app-side usage accounting or retry-related issue.

What steps can reproduce the bug?

This issue may not be fully reproducible by another user because it appears to be related to account-specific usage-limit accounting. However, the issue occurred during normal Codex app usage as follows:

1. Open the Codex app while signed in to a ChatGPT Pro account with 20x usage allowance.
2. Start a normal coding session in Codex.
3. Continue using Codex with a similar workload and intensity as previous sessions.
4. Observe the remaining usage allowance / limit status during the session.
5. Within approximately 1 hour, the allowance was depleted.
6. This appeared to consume roughly the amount of allowance that would normally last about 5 hours.
7. In previous similar Codex sessions, comparable usage usually lasted around 3 to 4 hours before reaching the limit.

No intentional change was made to my usage pattern that would explain such a large increase in consumption speed.

Relevant details:

• Session ID: [paste session ID here if available]
• Time of issue: [paste date and approximate time here]
• Codex App version: [paste version from About Codex here]
• Subscription: ChatGPT Pro with 20x usage allowance
• Platform: [paste platform output here]

Token / context details:

• Token limit usage: Not visible to me in the Codex app UI.
• Context window usage: Not visible to me in the Codex app UI.
• I could not find a detailed per-request usage breakdown showing why the allowance was consumed so quickly.

Code snippet:
Not applicable. This appears to be a usage accounting / limit tracking issue in the Codex app rather than a bug in project code. If there is a way to export diagnostic logs, session usage, token usage, or context-window usage from Codex, please let me know and I can provide them.

What is the expected behavior?

_No response_

Additional information

_No response_