List your company

Multi-step Reasoning on GSM-Hard

71.5AccuracyeMoT
13.67628.68843.758.712Jun 1, 2026

Evaluation results

MethodLinks
eMoTModels=Qwen-32B2026.0671.5
BoTModels=GPT-42026.0662
PaLModels=Codex (175B)2026.0661.2
ToTModels=GPT-42026.0624.4
Qwen-32B (Direct)Models=Qwen-32B2026.0615.9