List your company
ResearchFormerly WizWand

Mathematical Reasoning on AIME 25 (Acc, Tok)

71.3AccuracyOriginal
31.57241.88652.262.514May 9, 2026
Updated 5mo ago

Evaluation results

MethodLinks
OriginalModel Scale=Qwen3-4B-Thinking-2507, # Data=–2026.0571.320,973
Budget PromptModel Scale=Qwen3-4B-Thinking-2507, # Data=–2026.057120,362
SFT-S1.1Model Scale=DeepSeek-R1-Distill-Qwen-32B, # Data=1K2026.0564.813,983
HINT TUNINGModel Scale=Qwen3-4B-Thinking-2507, # Data=1K2026.0559.412,099
SFT-S1.1Model Scale=Qwen3-4B-Thinking-2507, # Data=1K2026.055612,273
SFT-S1.1Model Scale=DeepSeek-R1-Distill-Qwen-14B, # Data=1K2026.0553.314,104
OriginalModel Scale=DeepSeek-R1-Distill-Qwen-32B, # Data=–2026.0551.713,051
HINT TUNINGModel Scale=DeepSeek-R1-Distill-Qwen-32B, # Data=1K2026.05517,908
Budget PromptModel Scale=DeepSeek-R1-Distill-Qwen-32B, # Data=–2026.0547.511,982
OriginalModel Scale=DeepSeek-R1-Distill-Qwen-14B, # Data=–2026.0546.913,248
HINT TUNINGModel Scale=DeepSeek-R1-Distill-Qwen-14B, # Data=1K2026.0546.68,430
Budget PromptModel Scale=DeepSeek-R1-Distill-Qwen-14B, # Data=–2026.0545.612,339
SFT-S1.1Model Scale=DeepSeek-R1-Distill-Qwen-7B, # Data=1K2026.0539.415,319
Budget PromptModel Scale=DeepSeek-R1-Distill-Qwen-7B, # Data=–2026.053913,134
OriginalModel Scale=DeepSeek-R1-Distill-Qwen-7B, # Data=–2026.0537.914,576
HINT TUNINGModel Scale=DeepSeek-R1-Distill-Qwen-7B, # Data=1K2026.0537.79,144
AdaptThinkModel Scale=DeepSeek-R1-Distill-Qwen-7B, # Data=40K†2026.0536.912,063
EfficientRModel Scale=DeepSeek-R1-Distill-Qwen-7B, # Data=3.2K†, Alpha=0.12026.0535.411,902
VeriThinkerModel Scale=DeepSeek-R1-Distill-Qwen-7B, # Data=340K2026.0534.610,643
EfficientRModel Scale=DeepSeek-R1-Distill-Qwen-7B, # Data=3.2K†, Alpha=0.42026.0533.111,048