The International Mathematical Olympiad (IMO) is the world's most prestigious high school mathematics competition, held annually since 1959. Each year, over 600 students from more than 100 countries compete in a two-day, nine-hour exam featuring six extremely difficult problems in algebra, geometry, number theory, and combinatorics. Winning a gold medal at the IMO requires scoring among the top 12% of participants, typically solving five or six problems correctly. The prediction market question asks whether any artificial intelligence system will achieve a gold medal at the 2026 IMO, which would mark a major milestone in AI reasoning capabilities. This question is part of a broader effort to benchmark AI against human intellectual performance. The IMO Grand Challenge, launched in 2023, offers a $10 million prize for the first AI system that can win an IMO gold medal. The Artificial Intelligence Math Olympiad (AIMO) prize, funded by XTX Markets, provides a $5 million reward for similar achievement. These prizes have attracted teams from DeepMind, OpenAI, and various academic institutions. In 2024, Google DeepMind's AlphaGeometry solved 25 out of 30 IMO geometry problems from the past 20 years, performing at the level of a silver medalist in geometry but not across all problem types. Interest in this topic has grown rapidly because solving IMO problems requires genuine mathematical reasoning, not pattern matching or data retrieval. Unlike chess or Go, where AI has surpassed humans, mathematics involves constructing proofs and combining multiple logical steps. Many researchers consider IMO-level mathematics a key test for artificial general intelligence. The 2026 deadline creates urgency, as current systems still struggle with the full range of problem types, particularly combinatorics and number theory. The prediction market reflects a broader debate about the pace of AI progress. Optimists point to rapid improvements in large language models and specialized systems like AlphaGeometry. Skeptics note that no current AI can consistently solve even three of six IMO problems, and the gap between current performance and gold medal level remains large. The market price has fluctuated between 15% and 40% over the past year, reflecting genuine uncertainty about whether the necessary breakthroughs will occur by 2026.

Historical Context

The International Mathematical Olympiad began in 1959 in Romania with seven participating countries. The competition expanded rapidly during the Cold War, with the Soviet Union and United States dominating medal counts. By 2024, the IMO included 112 countries and 618 contestants. The maximum score is 42 points (7 per problem), and gold medals are awarded to the top 12% of participants. The cutoff for gold has ranged from 29 to 36 points in recent years, meaning a contestant typically needs to solve five of six problems correctly. AI's involvement with the IMO began informally in the 2010s when researchers tested early neural networks on past problems. In 2019, a system called 'Arqade' solved 2 of 20 geometry problems but used brute-force search rather than reasoning. The breakthrough came in 2021 when OpenAI's Codex, a language model trained on code, could solve about 10% of IMO problems when given as text. In 2022, DeepMind's system solved 4 of 30 problems from the 2021 IMO, but only after significant human engineering for each problem. The formal challenge structure emerged in 2023. The IMO Grand Challenge was announced in July 2023, offering $10 million for an AI that could win gold. XTX Markets launched the AIMO Prize in November 2023 with a $5 million purse. These prizes require AI to solve all six problems from a single IMO exam within a 12-hour window, with no human intervention allowed during the solving process. The first AIMO progress prize of $100,000 was awarded in 2024 to a team from Tsinghua University for solving 3 of 6 problems from the 2023 IMO.