I find claude models often use “tricks” like bash one liners, essentially excelling at surgical fixes. It does what i want more reliably, on smaller tasks.

GPT-5 can often be better at larger architectural changes, but i find that comes at the cost of instability/broken PRs. It often fails to capture intent or argues back, or just completely spirals out of control more often.

GPT-5 codex seemed to refuse valid requests like “make a change to break a test so we can test CI” (it over indexed on our agents.md and other instructions and then refused on the basis of “ethics” or some such)