In July 2025, AI systems reached the gold-medal standard at the International Mathematical Olympiad, one of the hardest tests of human mathematical reasoning. In a blog post dated July 21, 2025, Google DeepMind announced that an advanced version of Gemini with Deep Think “solved five of six International Mathematical Olympiad problems, earning 35 out of 42 possible points - a gold-medal performance.” DeepMind stressed that the result was “officially graded and certified by IMO coordinators using the same criteria as for student solutions,” and quoted the IMO President confirming the 35-point gold-medal score. It was an improvement over the previous year’s silver-medal-level result.
OpenAI separately reported that one of its models, evaluated on the same 2025 problems under contest-like conditions, also solved five of six problems for the same 35-point score, though as an unofficial evaluation rather than an IMO-certified one.
The significance is that a general-purpose reasoning system handled problems designed to stump the world’s most gifted high-school mathematicians - only a small fraction of human contestants earn gold. After AlphaGo’s mastery of Go, this extended machine achievement into open-ended mathematical proof, a domain long thought to require deep human creativity, and underscored how far reasoning models had advanced in under a year.