@busmark_w_nika I understand i think we have different usecases :)
Report
Last 2 months I use Codex more. I switch depending on the job. For a change that touches a real codebase, I care more about whether the agent can follow the existing patterns, run the tests, and leave a small reviewable diff than how pleasant the chat feels.
For open-ended exploration, I still want a more conversational loop. So I don’t think one tool wins every task. The deciding factor for me is how much verification the task needs after the first answer.
Replies
i am also thinjing about that but i think gpt 5.6 sol is the best rn, Keep telling us this news lik that
Murmell
@oflasfar I completely agree Codex > Claude
minimalist phone: reduce your screentime
I was more used to Claude Code. Codex is okay, but not good for chatting the way I liked :D
Murmell
@busmark_w_nika I understand i think we have different usecases :)
Last 2 months I use Codex more. I switch depending on the job. For a change that touches a real codebase, I care more about whether the agent can follow the existing patterns, run the tests, and leave a small reviewable diff than how pleasant the chat feels.
For open-ended exploration, I still want a more conversational loop. So I don’t think one tool wins every task. The deciding factor for me is how much verification the task needs after the first answer.