Codex seems to create a lot of useless functions
Resolved 💬 5 comments Opened Mar 25, 2026 by benglewis Closed Mar 25, 2026
💡 Likely answer: A maintainer (etraut-openai, contributor)
responded on this thread — see the highlighted reply below.
What issue are you seeing?
When coding, Codex seems to now frequently introduce functions which just pass their parameters and call another function and do nothing else - i.e. a useless function.
What steps can reproduce the bug?
There is no clear and easy way to reproduce the issue, but frequently Codex is introducing such functions. I see it when using GPT-5.4 . I do not know if other models are affected.
What is the expected behavior?
Functions which do nothing are not created and the _inner_ function is called directly instead.
Additional information
Codex thread ID for an example: 019d20cb-2d60-70b3-a95d-8734acf8a8c0
5 Comments
Thanks for the bug report. We'll share the feedback with the team responsible for training our codex models. We're tracking this feedback and similar reports so we can improve future models. And thanks for using the
/feedbackcommand. That additional information is useful.Another instance of this: 019d5009-2b32-7a90-817b-ff8e0ff68225 (codex thread id for an example)
This is happening a lot, to the point to which I am considering making using GPT-5.4 forbidden in our team. Is there anyone at OpenAI that we could work with to see if there are workarounds to fix this?
@benglewis, this is default model behavior, so there's not really any "fix" other than to train a new model. You can likely steer the model to avoid this behavior somewhat by adding explicit instructions to your AGENTS.md file.
From what I understand, GPT-5.5 is less likely to do this. I will update if I see it in GPT-5.5