DeepSeek's new AI model self-verifies Olympiad-level math problems
Chinese AI startup DeepSeek has released a new AI model designed to generate and self-verify mathematical theorems using advanced reasoning skills. The open-weight model, DeepSeek-Math-V2, achieved gold medal-worthy scores when tested on math problems from the 2025 International Mathematical Olympiad. It also scored 118 out of 120 on the Putnam 2024 math competition.