Recent progress from 2025, when OpenAI and Google DeepMind’s models first reached the gold-medal threshold of 35/42 on IMO problems under competition conditions, has shifted expectations for 2026. Frontier systems have continued scaling reasoning capabilities, with reported perfect 42/42 scores from models like Huawei’s Celia and others during the July 2026 Shanghai contest. However, trader consensus favoring “No” at 56% reflects caution around independent official verification by IMO graders, the difficulty of the remaining unsolved problem types, and strict market criteria requiring confirmed, reproducible results rather than self-reported benchmarks. Key near-term catalysts include final resolution announcements and any additional certified scores from major labs, which could quickly alter implied probabilities given the narrow margin.