List your company

Mathematical Reasoning on MATH 500 (Accuracy, Token count)

95.4AccuracyVanilla
85.62488.16290.793.238Apr 8, 2026

Evaluation results

MethodLinks
VanillaBackbone=Qwen3-14B2026.0495.44,503
DTSRBackbone=Qwen3-14B2026.04952,247
DTSRBackbone=Qwen3-32B2026.04952,660
VanillaBackbone=Qwen3-32B2026.0494.84,489
VanillaBackbone=Qwen3-8B2026.0494.45,067
DEERBackbone=Qwen3-14B2026.0494.42,601
DTSRBackbone=Qwen3-8B2026.04942,359
DEERBackbone=Qwen3-32B2026.04942,881
DEERBackbone=Qwen3-8B2026.0493.42,549
NoWAITBackbone=Qwen3-32B2026.0493.43,285
NoWAITBackbone=Qwen3-8B2026.0492.83,485
NoWAITBackbone=Qwen3-14B2026.0492.83,307
NoThinkingBackbone=Qwen3-14B2026.0488.2852
NoThinkingBackbone=Qwen3-32B2026.0486.4958
NoThinkingBackbone=Qwen3-8B2026.04861,067