A few months before the 2025 International Mathematical Olympiad (IMO) in July, a three-person team at OpenAI made a long bet that they could use the competition’s brutally tough problems to train an ...

OpenAI has launched FrontierScience, a new benchmark to assess expert-level AI scientific reasoning across physics, chemistry and biology, as models like GPT-5 increasingly support real research.

Scanning the crowd at a fancy soiree may reveal a wide array of neckties, each fastened with a highly complex mathematical object masquerading as fashion. An entire field of mathematics is devoted to ...