The post-training inference model SU-01 achieves gold medal performance in Olympiad-level questions

AIMPACT News, May 16 (UTC+8), a new paper proposes a systematic approach to convert post-training inference models into Olympic-level problem solvers and trains the SU-01 model based on this method.
The approach includes three steps: first, supervised fine-tuning using a reverse perplexity curriculum to instill strict proof search and self-checking behaviors;
then, extending these behaviors through two-stage reinforcement learning (transitioning from verifiable reward reinforcement learning to proof-level reinforcement learning);
finally, improving performance through scaling during testing.
The research team applied the method to a 30B-A3B backbone model, using approximately 340k sub-8K token trajectories for supervised fine-tuning, followed by 200 steps of reinforcement learning to obtain SU-01.
This model can perform stable reasoning on difficult problems, with trajectory lengths exceeding 100k tokens, achieving gold medal levels in competitions such as IMO 2025/USAMO 2026 and IPhO 2024/2025, and demonstrating generalization capabilities beyond mathematics and physics in scientific reasoning. (Source: InFoQ)