Perfect debugging score: Claude Sonnet 4.6 found and fixed all three bugs in a Python game test, outperforming its AI rivals. Mixed rival results: ChatGPT 5.5 identified two bugs but missed a key ...