# Competition-Level Mathematical Ability

OREAL-32B-SFT is a supervised fine-tuned model based on Qwen2.5-32B, specifically designed for mathematical reasoning tasks, serving as the initial policy model for the OREA reinforcement learning framework.