Codex works way better with open ai account than api key
Resolved 💬 6 comments Opened Aug 11, 2025 by tengocj Closed Oct 18, 2025
💡 Likely answer: A maintainer (pakrym-oai, contributor)
responded on this thread — see the highlighted reply below.
What version of Codex is running?
codex-cli 0.20.0
Which model were you using?
gpt-5
What platform is your computer?
Darwin 24.6.0 arm64 arm
What steps can reproduce the bug?
Just test logging in with your openai account with plus subscription and then use an api key to compare.
What is the expected behavior?
_No response_
What do you see instead?
_No response_
Additional information
Might be related to the api limits but was quite confusing ngl
6 Comments
Hi, can you please describe what works better?
For me it has constant rate limiting issues with an API key. I can't speak for the OP, though.
Yessir, when logging in with my account it's working better as when I am using my api key.
I did some research tho and it seems like that the api key limit is really low unless you pass p1 and get a higher limit - pretty sure it's related but need some confirmation to be sure.
@tibo-openai
@wojzaremba
@codex-maintainers
Since April, users have reported that Codex suddenly crashes when it exceeds organization-specific tokens-per-minute (TPM) limits (30,000 TPM for GPT‑5 for tier 1 organizations) without retrying or preserving session context.
I have now tracked multiple issues (e.g. #137, #157, #451, #690, #1395, and #2131) that consistently report this same critical failure spanning back to April: Codex CLI crashes when hitting rate limits instead of honoring retry-after instructions or preserving session context. Every time maintainers like @pakrym-oai have closed them without additional explanation, follow-up, or acknowledgment that the issue persists, despite continued user reports.
This pattern is deeply troubling and erodes trust. Trust not just in Codex CLI itself, but in the integrity of OpenAI’s support for its own developer tools. Could leadership clarify whether this is being actively addressed? Or what priorities or constraints are behind repeatedly closing unresolved issues? Confidence that the maintainers' decisions aren’t perceived as intentional dismissal or silent ignoring of legitimate user concerns is essential for developer's confidence in the platform.
We need explicit clarity and transparency on this.
We've made a number of improvements here
a) Higher throughput limits for API users
b) Better reconnection and recovery logic
c) Resume old sessions with
codex resumeGoing to close this, but please open a new issue if there are other things that hamper your productivity with codex, we want to make it delightful and are working hard towards this.
Thank you, @tibo-openai for finally getting this resolved! It appears to be now working on my end.