AI models of Google and OpenAI win milestone gold at global maths contest
Artificial Intelligence has reached a remarkable milestone. For the first time, AI models developed by Google DeepMind and OpenAI have performed at a gold-medal level in the International Mathematical Olympiad (IMO), a competition known for pushing the limits of human mathematical ability.
As reported by Business Standard, Google’s Gemini DeepThink and an experimental OpenAI system solved five out of six Olympiad-level problems, a score that placed them in the gold-medal bracket - a level typically achieved by only about 11% of human participants.
This is a significant breakthrough because Olympiad problems demand far more than simple calculation. They require creativity, deep reasoning, and multi-step proofs - qualities long thought to be uniquely human. The success of these models shows how far AI has advanced in tackling complex, abstract problems that were once considered beyond its reach.
Google’s model followed the same time constraints as the human contestants, working within the 4.5-hour limit, while OpenAI’s model demonstrated similar capability through independent evaluation. Both systems relied on advanced reasoning strategies and immense computational power to achieve their results.
This achievement isn’t just about winning a math competition; it hints at a future where AI plays an active role in research, engineering, and scientific discovery. While human insight remains essential, AI’s growing ability to reason and solve complex problems opens the door to powerful collaborations that could help tackle some of the toughest challenges in the world.