ResearchBenchmarksHuman Evaluation on Management subsetFollow93Win RateMENTOR85.7287.6189.591.39Nov 10, 2025Evaluation ResultsMethodMethodLinksWin RateTie RateLoss RateMENTORBackbone=DeepSeek-R1Backbone=DeepSeek-R12025.119352MENTORBackbone=Qwen3-235BBackbone=Qwen3-235B2025.118974MENTORBackbone=GPT-4oBackbone=GPT-4o2025.1186131