List your company

Mathematical Reasoning on MATH 500 (Accuracy, Think ↓, Route%)

92.6AccuracyDART
75.9680.2884.688.92Jun 22, 2026

Evaluation results

MethodLinks
DARTBackbone=DeepSeek-V3.2, Hosted API=true2026.0692.656—
SC-RouteBackbone=DeepSeek-V3.2, Hosted API=true2026.0690.6——
SC-RouteBackbone=Qwen3-32B2026.0688.5——
DARTBackbone=Qwen3-32B2026.0688.56980
ATBackbone=DeepSeek-V3.2, Hosted API=true2026.0688.4——
DARTBackbone=Qwen3-8B2026.0688.26778
SC-RouteBackbone=Qwen3-8B2026.0687.6——
ATBackbone=Qwen3-14B2026.0687.6——
SC-RouteBackbone=Qwen3-14B2026.0687.6——
DARTBackbone=Qwen3-14B2026.0687.63783.4
ATBackbone=Qwen3-32B2026.0686.2——
ATBackbone=Qwen3-8B2026.0685.6——
NTBackbone=DeepSeek-V3.2, Hosted API=true2026.0684.8——
NTBackbone=Qwen3-32B2026.0682.2——
NTBackbone=Qwen3-14B2026.0681.2——
NTBackbone=Qwen3-8B2026.0676.6——