IME it works even better if you use another model for review. We've seen code by cc and review by gpt5.2/3 work very well. Also works with planning before any

IME it works even better if you use another model for review. We've seen code by cc and review by gpt5.2/3 work very well. Also works with planning before any coding sessions. Gemini + Opus + GPT-xhigh works to get a lot of questions answered before coding starts.