**Frontier AI systems have already demonstrated gold-medal and perfect scores on the 2026 IMO problems through independent and some officially graded evaluations, yet the market prices "No" at 96.4% because resolution criteria emphasize formal participation in the live human contest rather than benchmark performance alone.** Multiple models, including Chinese systems like RedNote’s dots-note-3.0 and Huawei’s Celia, reached 42/42 on the July 2026 Shanghai problems, while others such as Claude Opus 5 and various GPT variants matched or exceeded the gold threshold in natural-language proof settings. These results reflect rapid capability gains from 2025 gold-level benchmarks, but strict eligibility rules limiting entrants to human students under 20, combined with the IMO’s emphasis on in-person competition without external tools, have kept AI entries from qualifying. Traders assign low odds to last-minute rule changes or regulatory shifts that would enable official AI participation before year-end.