List your company

Mathematical Reasoning on Average (AIME24, AIME25, AMC23, MATH500, OlympiadBench)

61.03Pass@1MEL
35.352442.018748.68555.3513Feb 10, 2026

Evaluation results

MethodLinks
MELBackbone=Qwen3-14B-Base2026.0261.0360.0774.94
GRPOBackbone=Qwen3-14B-Base2026.0256.356.4773.24
MELBackbone=Qwen3-8B-Base2026.0252.7952.2771.17
GRPOBackbone=Qwen3-8B-Base2026.0248.4351.0767.33
MELBackbone=Qwen3-4B-Base2026.0245.4846.2563.41
GRPOBackbone=Qwen3-4B-Base2026.0241.5644.9261.04
BaselineBackbone=Qwen3-8B-Base2026.0241.338.0660.6
BaselineBackbone=Qwen3-14B-Base2026.0241.2137.2662.34
BaselineBackbone=Qwen3-4B-Base2026.0236.3432.0655.16