I tested it yesterday. It is pretty bad. Just like with Composer 2, it's fast, but quality is nowhere near what Cursor claims with their benchmarks. It is not even at Opus 4.5 level.

I gave it a mix of refactoring tasks and new feature tasks. For each one, I had it write a plan, then I had Codex review it. Codex found major issues with every plan: patterns that don't match the rest of the code base, hallucinated variable/function names, and even outright bugs in the way the plan was written. I fed the feedback to Composer 2. After it made the changes and implemented the revised plan, I had Codex and Opus 4.7 do code reviews, and once again both of them found major bugs.

Overall it was a very frustrating experience. I feel like I wasted a whole day. Which is sad, as I have been looking for an excuse to come back to Cursor. But as things stand, Codex + CC combo cannot be beat, not just in terms of price but also quality.