Achieving Gold-Medal-Level Olympiad Reasoning via Simple and Unified Scaling

Benchmark Model Rank Results
mathematical-reasoning-on-amo-benchSU-01#2Accuracy: 59.8
science-question-answering-on-frontierscience-researchSU-01#4Accuracy: 11.7
science-question-answering-on-frontierscience-olympiadSU-01#6Accuracy: 61.5
mathematical-reasoning-on-aime-2025SU-01#10Accuracy: 94.6