It's been shockingly bad for me - for another example when asked to make a new python script building off an existing one; for some cursed reason the model choose to .read() the py files, use 100 of lines of regex to try to patch the changes in, and exec'd everything at the end...

I generally think codex is doing well until I come in with my Opus sweep to clean it up. Claude just codes closer to the way my brain works. codex is great at finding numerical stability issues though and increasingly I like that it waits for an explicit push to start working. But talking to Claude Code the way I learned to talk to codex seems to work also so I think a lot of it is just learning curve (for me).