List your company

Mathematical Reasoning on AIME 2024 (Accuracy and Token Costs)

40Sample CountSC
17.95223.67629.435.124Feb 10, 2026

Evaluation results

MethodLinks
SCBackbone=Qwen2.5-3B2026.0240—42.30.112
SCBackbone=Qwen2.5-7B2026.0240—42.20.1667
SCBackbone=Gemma3-4B2026.0240—163.80.1333
ESCBackbone=Qwen2.5-3B2026.0238.81—430.112
DSCBackbone=Qwen2.5-3B2026.0237.854.941.30.112
ESCBackbone=Qwen2.5-7B2026.0237.11—41.50.1667
ACBackbone=Qwen2.5-3B2026.0234.68—44.10.112
DSCBackbone=Qwen2.5-7B2026.0234.11537.80.1667
ESCBackbone=Gemma3-4B2026.0233.7—139.30.1333
ACBackbone=Qwen2.5-7B2026.0232.3—41.20.1667
DSCBackbone=Gemma3-4B2026.0229.7817122.50.1333
ACTSCBackbone=Qwen2.5-3B2026.0228.17—33.70.112
ACBackbone=Gemma3-4B2026.0227.3—116.70.1333
ACTSCBackbone=Qwen2.5-7B2026.0224.7—290.1667
ACTSCBackbone=Gemma3-4B2026.0218.8—78.10.1