Of course if you give codex extra help by using a command/plugin then it’s going to outperform a claude agent that has no extra help. Funny how when asked if he tried to use the same command/plugin with claude code, he stated that he didn’t have time to. The OpenAI engineer clearly just wants to slander claude. OpenAI people are always doing this kind of crap. I’m going to try to see if I can do a local performance comparison when I have the chance this week. I’ll intentionally try to give a vague feature-request prompt to both claude and codex, no extra plugins/commands allowed, and see which one produces more bugs. I’ll edit this post on the weekend. Unfortunately, I don’t have a Mac anymore, so I can’t even test the performance of claude with the plugin vs codex with the command, so that don’t expect that from me (sorry).
Plus, has anyone tried codex 5.1? I want to be respectful and fair here… but respectfully…it’s complete garbage compared to Opus.
I want to call bullshit on this.
Of course if you give codex extra help by using a command/plugin then it’s going to outperform a claude agent that has no extra help. Funny how when asked if he tried to use the same command/plugin with claude code, he stated that he didn’t have time to. The OpenAI engineer clearly just wants to slander claude. OpenAI people are always doing this kind of crap. I’m going to try to see if I can do a local performance comparison when I have the chance this week. I’ll intentionally try to give a vague feature-request prompt to both claude and codex, no extra plugins/commands allowed, and see which one produces more bugs. I’ll edit this post on the weekend. Unfortunately, I don’t have a Mac anymore, so I can’t even test the performance of claude with the plugin vs codex with the command, so that don’t expect that from me (sorry).
Plus, has anyone tried codex 5.1? I want to be respectful and fair here… but respectfully…it’s complete garbage compared to Opus.