List your company

Mathematical Reasoning on MATH500 (pass@1 accuracy)

82.2Pass@1 AccuracyAERO
67.6471.4275.278.98Feb 3, 2026

Evaluation results

MethodLinks
AEROBackbone Model=Qwen3-8B-Base, Evolutionary Round=R52026.0282.2
R-ZeroBackbone Model=Qwen3-8B-Base2026.0282
AEROBackbone Model=Qwen3-8B-Base, Evolutionary Round=R42026.0281.8
AEROBackbone Model=Qwen3-8B-Base, Evolutionary Round=R32026.0279.8
AEROBackbone Model=Qwen3-8B-Base, Evolutionary Round=R12026.0279.4
AEROBackbone Model=Qwen3-8B-Base, Evolutionary Round=R22026.0279.2
Qwen3-8B-BaseBackbone Model=Qwen3-8B-Base2026.0278
Absolute ZeroBackbone Model=Qwen3-8B-Base2026.0276.6
Absolute ZeroBackbone Model=Qwen3-4B-Base2026.0276.2
R-ZeroBackbone Model=Qwen3-4B-Base2026.0274.8
AEROBackbone Model=Qwen3-4B-Base, Evolutionary Round=R52026.0274.8
AEROBackbone Model=Qwen3-4B-Base, Evolutionary Round=R42026.0274.4
AEROBackbone Model=Qwen3-4B-Base, Evolutionary Round=R32026.0273.8
AEROBackbone Model=Qwen3-4B-Base, Evolutionary Round=R22026.0272.6
AEROBackbone Model=Qwen3-4B-Base, Evolutionary Round=R12026.0270.8
Qwen3-4B-BaseBackbone Model=Qwen3-4B-Base2026.0268.2