Simple git commit and push consumed 9% of my 5-hour limit on Plus

Open 💬 1 comment Opened May 3, 2026 by squarepots

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

26.430.10722

What subscription do you have?

ChatGPT Plus

What platform is your computer?

Microsoft Windows NT 10.0.26200.0 x64

What issue are you seeing?

I am using Codex with GPT-5.5 High on a ChatGPT Plus subscription.

A very small task — asking Codex to perform a git commit and push — consumed 9% of my 5-hour usage limit.

This seems wrong or at least extremely disproportionate. The task did not involve a large code generation session, long multi-file refactor, or heavy reasoning. It was basically a routine git operation.

I understand usage may be token-based, but consuming 9% of the 5-hour limit for a commit and push workflow makes the displayed limit feel inaccurate or the accounting unexpectedly expensive.

Please investigate whether the usage accounting is correct for small operational tasks like git commit/push, especially when using GPT-5.5 High.

Session ID:
[Paste session id]

Token limit usage:
[Paste token usage if available]

Context window usage:
[Paste context window usage if available]

What steps can reproduce the bug?

  1. Open Codex App.
  2. Use GPT-5.5 High.
  3. Work in an existing git repository.
  4. Ask Codex to commit and push the current changes.
  5. Observe the usage meter / 5-hour limit consumption.
  6. In my case, this simple git commit and push consumed about 9% of the 5-hour limit.

No special code snippet is required. The issue is about usage accounting for a simple git operation.

What is the expected behavior?

A simple git commit and push should consume only a very small amount of the Plus usage limit.

If usage is token-based, the UI should make that clear and the accounting should still be proportional to the actual work performed. A routine commit/push operation should not consume around 9% of a 5-hour limit unless there is a clear explanation, such as extremely large context, hidden output, repeated retries, or unexpectedly high token usage.

Additional information

This makes the Plus limit difficult to reason about. If a small git operation can consume 9% of the 5-hour limit, then the practical value of the limit is much lower than expected.

It would be helpful if Codex showed a more detailed usage breakdown, for example:

  • input tokens
  • output tokens
  • cached tokens
  • context window size
  • tool/runtime overhead
  • retries or hidden operations
  • whether git commands or terminal output contributed significantly to usage

This would make it easier to tell whether the usage is expected or whether the app is overcounting.

View original on GitHub ↗

This issue has 1 comment on GitHub. Read the full discussion on GitHub ↗