AI Reaches Gold Medal Level at Math Olympiad

Did you know that an artificial intelligence has reached the gold medal level at the International Mathematical Olympiad? In two thousand twenty-five, a search version of Gemini Deep Think solves five out of six problems. Here, you have to produce a reasoning, not just guess a result. These problems require building a demonstration that proofreaders can examine. A geometric figure illustrates this idea, without claiming to be a solution of the contest. A pretty figure proves nothing alone: each step must be justified to become a true mathematical proof.

The system scores thirty-five points out of forty-two, with an official evaluation announced. This score reaches the gold threshold of this edition. It is not necessary to invent a physical medal, a place of student in the room or a victory over all mathematicians.

The result is an advanced search version and specific problems. It does not guarantee that every response from a mainstream chatbot is fair. The progress is spectacular, but the demand remains the same: a demonstration is true, even when a machine seems convincing.