High Schoolers Outperform AI in International Math Olympiad

Published: July 28, 2025

In a remarkable display of human intellect, 26 high school students surpassed the performance of advanced AI models in the 2025 International Mathematical Olympiad (IMO), a premier global mathematics competition. Notably, AI systems from OpenAI and Google DeepMind achieved gold medal scores, but these students outperformed them, scoring higher on average. The Wall Street Journal, 1LinkedIn

The IMO, held annually, presents six challenging problems over two days. While AI models like OpenAI's o3 and DeepMind's AlphaGeometry2 have made significant strides in mathematical reasoning, they still fall short of human capabilities in certain areas. For instance, AlphaGeometry2, an AI developed by DeepMind, solved 42 out of 50 geometry problems from previous IMO competitions, surpassing the average gold medalist score of 40.9 .arXiv

Despite these advancements, the performance of human students remains exceptional. Their ability to apply creative problem-solving techniques and adapt to novel challenges continues to outpace current AI models. This achievement underscores the enduring value of human ingenuity in complex problem-solving scenarios.

As AI continues to evolve, this event serves as a reminder of the unique strengths of human cognition. While AI can assist in various domains, the capacity for innovative thinking and adaptability remains a distinct human trait.