List your company

Mathematical Reasoning on MATH 500 (Acc, Tok, CR)

95.2AccuracyVanilla
79.80883.80487.891.796Apr 19, 2026

Evaluation results

MethodLinks
VanillaBackbone=Qwen3-8B2026.0495.25,138100
TRACEBackbone=Qwen3-8B2026.0494.13,23162.9
DynasorBackbone=Qwen3-8B2026.0493.73,87375.4
VanillaBackbone=Qwen3-4B2026.0493.65,229100
TRACEBackbone=Qwen3-4B2026.0493.23,78872.4
TALEBackbone=Qwen3-8B2026.0493.13,54469
DEERBackbone=Qwen3-8B2026.0492.93,46867.5
DEERBackbone=Qwen3-4B2026.0492.83,94475.4
VanillaBackbone=Gemini2.5-Flash2026.0492.81,449100
TRACEBackbone=Gemini2.5-Flash2026.0492.21,16680.5
DynasorBackbone=Gemini2.5-Flash2026.04921,31190.5
TALEBackbone=Gemini2.5-Flash2026.0491.91,41197.4
DynasorBackbone=Qwen3-4B2026.0491.44,33983
DEERBackbone=Gemini2.5-Flash2026.0491.21,22484.5
NoThinkBackbone=Gemini2.5-Flash2026.0490.61,40096.6
TALEBackbone=Qwen3-4B2026.0489.72,93856.2
VanillaBackbone=R1-Distilled-Llama-8B2026.0487.22,934100
DynasorBackbone=R1-Distilled-Llama-8B2026.0486.42,69391.8
DEERBackbone=R1-Distilled-Llama-8B2026.0486.42,71892.6
TRACEBackbone=R1-Distilled-Llama-8B2026.0485.42,43683
TALEBackbone=R1-Distilled-Llama-8B2026.0485.32,45583.7
NoThinkBackbone=R1-Distilled-Llama-8B2026.0485.23,024103
NoThinkBackbone=Qwen3-8B2026.0484.292318
NoThinkBackbone=Qwen3-4B2026.0480.490817.4