live model showdown

Ran the same code review through claude opus and gpt 4 in live comparison. The differences were obvious — and the better response wasn’t the one I expected.