You should still compare the outputs for a couple of reasons
1. When GPT Pro has to navigate the github it spends a ton of context on tool calls and navigation which burns it's limited context window
2. It likely wont read as much context as if you prep context for it ahead of time
3. Idk if you tried the codereviews flow with the export tool from the agent, but you can get insanely deep code review prompts packaged up for pro
Still - do what works best for you!