List your company

Step-wise Verification on ProcessBench Omni-MATH

63.1TNRCons@4
5.79620.67335.5550.427May 20, 2026

Evaluation results

MethodLinks
Cons@4Backbone=Qwen3-8B2026.0563.171.467
VerifySteer-UniBackbone=Qwen3-8B2026.0561.976.868.5
VerifySteer-BiBackbone=Qwen3-8B2026.0561.877.268.6
Qwen3-14BBackbone=Qwen3-14B2026.0559.680.268.4
Prompt Opt.Backbone=Qwen3-8B2026.0558.675.566
Basic PromptBackbone=Qwen3-8B2026.0553.873.462.1
Qwen2.5-72BBackbone=Qwen2.5-72B2026.0537.583.451.8
Llama-3.3-70BBackbone=Llama-3.3-70B2026.05298843.6
VerifySteer-UniBackbone=Qwen3-1.7B2026.0512.659.320.9
VerifySteer-BiBackbone=Qwen3-1.7B2026.0512.664.321.1
Cons@4Backbone=Qwen3-1.7B2026.051262.220
Prompt Opt.Backbone=Qwen3-1.7B2026.0510.457.317.6
Basic PromptBackbone=Qwen3-1.7B2026.05882.614.7