List your company

ResearchFormerly WizWand

Mathematical Reasoning Process Evaluation

Benchmarks

DatasetSOTA methodMetricTrendResultsLast updated

ProcessBench (test)

Qwen2.5-Math-7B-MathShepherd-r

96.9GSM8K Accuracy

35

May 20, 2026

Llama-3.3-70B-Instruct

82.9GSM8K Accuracy

28

Feb 26, 2026