Codex model can't be used for reviews
Resolved 💬 4 comments Opened Nov 22, 2025 by matssk Closed Dec 1, 2025
💡 Likely answer: A maintainer (github-actions[bot], contributor)
responded on this thread — see the highlighted reply below.
What version of Codex is running?
0.63.0
What subscription do you have?
Pro
Which model were you using?
gpt-5.1-codex-max
What platform is your computer?
_No response_
What issue are you seeing?
Trying to use the GPT 5.1 codex max for reviews using the /review command, it does not find bugs.
After repeatedly running /review with GPT 5.1 Codex Max I tried the same with GPT-5.1 and each time it found a new bug, while GPT 5.1 Codex Max was unable to find any of them.
What steps can reproduce the bug?
Type /review when you have code with bugs using the GPT 5.1 Codex Max model
It does not find any bugs, even when run many times.
What is the expected behavior?
It should find bugs, like the GPT-5 or 5.1
Additional information
_No response_
4 Comments
Potential duplicates detected. Please review them and close your issue if it is a duplicate.
Powered by Codex Action
/review function always use
gpt-5.1-codexwith medium reasoning effort ifreview_modelis not set inconfig.tomlHm. If that's so this might just have been a coincidence then. I tried running /review in multiple session with changing model using /model and I ran it many times this way with model set to gpt-5.1 and gpt-5.1-codex-max. When I set the model to codex max it didn't find any of the bugs, but when it was on gpt-5.1 it found them. Now if /model doesn't actually change the model used for the review though, it must have just been the random seeds for when I had set model to gpt-5.1, just happened to be randomly "better" for finding the bugs.
Thanks for the bug report. We'll share the feedback with the team responsible for training our codex models.