Claude Opus 4.7 outperformed ChatGPT-5.5 in structured reasoning tests, including a probability challenge where it provided both the correct answer and a generalized formula. The release also ...