List your company

Mathematical Reasoning on AIME 2025 (Weighted majority-voting accuracy, Pass@128, Pass@1)

69.32Weighted Majority Voting AccuracyOTV
24.308835.994447.6859.3656Mar 1, 2026

Evaluation results

MethodLinks
OTVBackbone=QWEN3-4B2026.0369.32——
DeepConfBackbone=QWEN3-4B2026.0366.77——
GenRMBackbone=QWEN3-4B2026.0366.72——
VersaPRM-8BBackbone=QWEN3-4B2026.0366.72——
Math-Shepherd-7BBackbone=QWEN3-4B2026.0366.56——
MAJ@128Backbone=QWEN3-4B2026.0366.46——
Qwen2.5-PRM800K-7BBackbone=QWEN3-4B2026.0366.35——
Qwen2.5-PRM-7BBackbone=QWEN3-4B2026.0363.49——
AceMath-RM-7BBackbone=QWEN3-4B2026.0360——
OTVBackbone=DAPO-QWEN-32B2026.0349.58——
Qwen2.5-PRM800K-7BBackbone=DAPO-QWEN-32B2026.0348.18——
Qwen2.5-PRM-7BBackbone=DAPO-QWEN-32B2026.0346.77——
DeepConfBackbone=DAPO-QWEN-32B2026.0344.79——
Math-Shepherd-7BBackbone=DAPO-QWEN-32B2026.0343.65——
VersaPRM-8BBackbone=DAPO-QWEN-32B2026.0343.18——
AceMath-RM-7BBackbone=DAPO-QWEN-32B2026.0342.76——
GenRMBackbone=DAPO-QWEN-32B2026.0342.55——
MAJ@128Backbone=DAPO-QWEN-32B2026.0341.77——
OTVBackbone=QWEN3-8B2026.0333.85——
GenRMBackbone=QWEN3-8B2026.0332.66——
DeepConfBackbone=QWEN3-8B2026.0331.87——
Qwen2.5-PRM-7BBackbone=QWEN3-8B2026.0329.48——
Qwen2.5-PRM800K-7BBackbone=QWEN3-8B2026.0329.32——
VersaPRM-8BBackbone=QWEN3-8B2026.0329.11——
MAJ@128Backbone=QWEN3-8B2026.0328.44——
AceMath-RM-7BBackbone=QWEN3-8B2026.0326.72——
Math-Shepherd-7BBackbone=QWEN3-8B2026.0326.04——
Pass@1Backbone=QWEN3-4B2026.03——46.67
Pass@1Backbone=QWEN3-8B2026.03——19.32
Pass@1Backbone=DAPO-QWEN-32B2026.03——36.42
Pass@128Backbone=QWEN3-4B2026.03—83.13—
Pass@128Backbone=QWEN3-8B2026.03—56.98—
Pass@128Backbone=DAPO-QWEN-32B2026.03—68.75—