OpenAI has claimed gold-level performance in an evaluation modeled after the 2025 International Math Olympiad (IMO), testing its “experimental general reasoning LLM” on the same problem statements used in the official human competition.

The unnamed model was tested under standard IMO conditions, solving problems across two 4.5-hour exams without tools or internet. OpenAI claims the model solved 5 of 6 problems, scoring 35/42 — enough for a gold medal. Each solution was graded by three former IMO medalists, with unanimous consensus required for scoring.

However, Google DeepMind has rebuked the claim, citing the IMO’s internal marking guidelines and asserting that no official claim can be made without formal adherence.

While the achievement may draw scrutiny, it represents a significant milestone in AI reasoning. With the model still unreleased, it’s clear OpenAI has more advancements in the pipeline.