OpenAI says its new model, GPT-5.1, is much better at math and coding tests like AIME 2025 and Codeforces. It can explain ...