List your company

Arithmetic Reasoning on MultiArith (Source/Unseen Library Accuracy)

99.64Source Library AccuracySIGMA
93.888895.381996.87598.3681Jun 18, 2026

Evaluation results

MethodLinks
SIGMABase Model=GPT-4o-mini2026.0699.6499.460.18
CARDBase Model=GPT-4o-mini2026.0696.9495.950.99
G-DesignerBase Model=GPT-4o-mini2026.0696.7595.51.25
GPTSwarmBase Model=GPT-4o-mini2026.0696.0793.932.14
LLM-DebateBase Model=GPT-4o-mini2026.0694.1191.962.15