gold-medal standard
Performance level meeting the criteria for a gold medal at the International Mathematical Olympiad (IMO), typically requiring top 5% problem-solving accuracy among competitors.
- deepmind’s gemini-25-pro with deep-think achieved this standard at the International Mathematical Olympiad (IMO) in 2026.
- Represents breakthrough in AI mathematical reasoning, matching human gold-medal performance (top 5%).
- Compared to previous AI attempts and concurrent openai achievement in mathematical reasoning.
2026 04 14 Gemini deep think sam witteveen