Humans Still Lead as AI Scores Gold at 2025 Math Olympiad

At the 2025 International Mathematical Olympiad (IMO), held in Queensland, Australia, artificial intelligence made history—but it didn’t take the spotlight away from human brilliance.

For the first time ever, AI models developed by Google and OpenAI achieved gold-level scores at the world’s most prestigious high school math competition. Both systems scored 35 out of a possible 42 points—an impressive feat that marks a major leap from previous years.

But despite the achievement, neither model reached perfection. That honor went to five human contestants who managed to earn full marks.

This year’s IMO featured six advanced mathematical problems, with students from around the world—aged 20 and below—given just four and a half hours to solve them. Google’s Gemini and OpenAI’s reasoning model matched the gold medal threshold, but fell one step short of the top.

According to Google, Gemini successfully solved five out of six problems, a big jump from last year’s performance where it only managed a silver score. “We can confirm that Google DeepMind has reached the much-desired milestone,” said IMO president Gregor Dolinar. “Their solutions were astonishing—clear, precise, and easy to follow.”

OpenAI also shared its results, revealing that their model was tested under the same conditions as human contestants. Each solution was independently graded by three former IMO medalists.

Google further noted that unlike last year—when solving the problems took days of computation—Gemini completed its solutions within the official competition time window, showing just how far the technology has progressed.

Still, despite the breakthroughs, the organizers were careful to point out a limitation. They couldn’t fully verify the computing resources used by the AI systems or confirm if human help was involved during the process.

Dolinar acknowledged the significance of the achievement. “It’s exciting to see how far AI has come in mathematics,” he said. “But we must also recognize the value of human insight—the ability to think beyond rules, apply intuition, and solve creatively under pressure.”

In the end, while AI models proved they can keep up with some of the world’s brightest young minds, they still haven’t surpassed them. At least, not yet.