Codex run out of TPM very quickly
Resolved 💬 9 comments Opened Sep 3, 2025 by damtt Closed Sep 4, 2025
💡 Likely answer: A maintainer (pakrym-oai, contributor)
responded on this thread — see the highlighted reply below.
What version of Codex is running?
0.27.0
Which model were you using?
gpt-5 medium (default)
What platform is your computer?
Darwin 24.5.0 arm64 arm
What steps can reproduce the bug?
- Launch
codex - Prompt a simple task like
Change the default font size for the app to 14px, except some components. - Wait for it to generate a checklist and perform file scanning.
- Run out of TPM with error:
stream error: stream disconnected before completion: Rate limit reached.....
What is the expected behavior?
_No response_
What do you see instead?
<img width="2070" height="1130" alt="Image" src="https://github.com/user-attachments/assets/7324ad4b-68b7-4c19-9f0c-dc5890caecd3" />
Additional information
Fresh setup NextJS project with less than 10 components, using Typescript, Tailwindcss, Shadcn UI.
9 Comments
Are you logged in via an API key? Can you please run
/status?I'm logged in using my API key on Tier 1 with 30k TPM for the GPT-5 model :|
30k TPM is a pretty low limit and the context size reaches 30k very quickly. I recommend trying codex with a Plus or Pro accounts that have much more generous limits.
It's been acting weird lately, I believe since the 0.27 update, because it doesn't "eat" as many tokens as before... at least before the 0.27 upgrade, it can handle small tasks like generating a unit test or implementing a simple feature.
I also have a Pro account, but it reached the limit, and I can't use it until next week :|
Can you please try 0.29? There was a regression in retrying token limit errors that we fixed yesterday.
I tried version 0.29 after posting this, but I still have the same issue.
Yes still a huge problem. I put in $100 9 days ago and am still limited to Tier 1. So not sure what I am meant to do?
Duplicate of #2629