List your company

Mathematical Reasoning on OlympiadBench (Mean@8)

60.1Mean Score @8FlowRL
22.45232.2264251.774May 17, 2026

Evaluation results

MethodLinks
FlowRLBackbone=Qwen3-4B-Base2026.0560.1
GSPOBackbone=Qwen3-4B-Base2026.0557.2
DISABackbone=Qwen3-4B-Base2026.0556.8
GRPOBackbone=Qwen3-4B-Base2026.0547.5
FlowRLBackbone=Qwen2.5-7B2026.0532.4
DISABackbone=Qwen2.5-7B2026.0531.9
VanillaBackbone=Qwen3-4B-Base2026.0530.1
GRPOBackbone=Qwen2.5-7B2026.0529.1
GSPOBackbone=Qwen2.5-7B2026.0528.8
VanillaBackbone=Qwen2.5-7B2026.0523.9