Title: Unexpected high credit consumption and possible rate limit issue with Codex 5.4

Resolved 💬 2 comments Opened Mar 19, 2026 by luanpgmachado Closed Mar 19, 2026
💡 Likely answer: A maintainer (github-actions[bot], contributor) responded on this thread — see the highlighted reply below.

Hello,

I'm reporting what appears to be abnormal credit consumption behavior when using Codex 5.4.

Today, right after my plan reset, I started a session involving a relatively small and structured task (Figma template generation based on an HTML mockup and documentation). Within a very short period of time, my weekly usage jumped to approximately 34% consumed (only 66% remaining).

This does not align with my previous experience using earlier models for the exact same type of task, where the consumption was significantly lower (well under 1–5%).

Key details:

  • Task type: structured UI/template generation with multiple documents (HTML + markdown specs)
  • No excessive iterations or loops beyond normal usage
  • Single session consumed a disproportionately high percentage of the weekly limit
  • Context size reached ~70k tokens, but this was not unusual compared to prior usage patterns

Observations:

  • Credit usage appears inconsistent with previous model behavior
  • Consumption per request seems significantly higher than expected
  • Similar reports have been observed recently from other users

Request:

Could you please:

  1. Investigate whether this behavior is expected or if there is a miscalculation or rate limit issue
  2. Review my usage for this session
  3. Consider resetting or partially restoring my consumed credits if this is confirmed to be unintended behavior

I rely on this tool for development workflows, and this level of consumption makes it difficult to use reliably.

Thank you.

View original on GitHub ↗

2 Comments

github-actions[bot] contributor · 5 months ago

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

  • #15039
  • #14628
  • #14593
  • #14752

Powered by Codex Action

etraut-openai contributor · 5 months ago

Please see #14593.