In 2025, for the first time, AI models from Google DeepMind and OpenAI earned gold medals in the International Math Olympiad. The event featured two 4.5-hour exams with six problems, where the AI models solved five of the six challenges within the allotted time. Both models scored 35 points out of 42, equating to the gold medal cutoff. Google DeepMind collaborated with IMO for grading, while OpenAI used former IMO medalists for evaluation. The achievement represents a significant milestone in AI's capabilities in mathematical problem-solving, surpassing Google DeepMind's prior silver medal performance.
Google DeepMind and OpenAI's AI models achieved gold medal status at the 2025 International Math Olympiad, solving five out of six problems without human intervention.
Both AI models scored 35 points out of a possible 42 points, marking the first time AI has reached such a level of success at this competition.
Collection
[
|
...
]