Capability Ledger · verified · Mathematics
In July 2025 OpenAI reported that an experimental reasoning model scored 35/42 on the IMO 2025 paper under contest-like constraints, graded by former medalists.
Capability score
1.49
Does not count toward Net Good or the leaderboard.
Independent gold-threshold IMO performance is a major capability milestone (tier 3), not Net Good. It is not shared with Gemini. Attribution is high. Evidence is journalism and developer posts rather than official IMO grading (0.50).
OpenAI did not enter IMO 2025 as a contestant. An experimental reasoning model produced natural-language solutions to five of six problems for 35 of 42 points, scored by three former medalists under contest-style time limits. Gemini Deep Think independently reached the same score with official IMO grading; that is a separate event. This is a capability demonstration on a contest paper whose answers already existed, not newly created mathematics, and does not count toward Net Good.
Experimental reasoning model scored 35/42 by former IMO medalists on the 2025 paper.
Independent parallel result; not official IMO grading and not a collaboration with Gemini.
Attribution 0.8500 · Credit share 100% · OpenAI
An OpenAI experimental model scored 35/42 on IMO 2025 under contest-like constraints.
outcome · supported
OpenAI officially won an IMO gold medal as a contestant.
significance · disputed
independent sources
Scientific American · journalism · 21 Jul 2025
Reuters · journalism · 21 Jul 2025
Secondary domains: Computer Science
14 Sep 2026 · 0.00 → 1.49
Imported events under methodology 0.2.
