List your company

Mathematical Reasoning on BeyondAIME (pass@1, pass@64)

8.3Pass@1CurveRL
0.8122.7564.76.644May 23, 2026

Evaluation results

MethodLinks
CurveRLBackbone=Qwen3-4B-Base2026.058.340.4
MaxRLBackbone=Qwen3-4B-Base2026.058.232.5
GRPOBackbone=Qwen3-4B-Base2026.05829
BaseBackbone=Qwen3-4B-Base2026.05423.8
CurveRLBackbone=Qwen3-1.7B-Base2026.051.423.6
MaxRLBackbone=Qwen3-1.7B-Base2026.051.321.8
BaseBackbone=Qwen3-1.7B-Base2026.051.116.9
GRPOBackbone=Qwen3-1.7B-Base2026.051.115.6