List your company

SQL Agent on WikiSQL

60.8AccuracyReflection
26.16835.15944.1553.141May 27, 2026

Evaluation results

MethodLinks
ReflectionInference Method=MCTS2026.0560.8
ReflectionInference Method=Best-of-N (pass@5), Temperature=0.92026.0558.8
No MemoryInference Method=Best-of-N (pass@5), Temperature=0.92026.0549
No MemoryInference Method=MCTS2026.0549
LiTS-FactInference Method=Best-of-N (pass@5), Temperature=0.92026.0547.1
Raw SiblingInference Method=MCTS2026.0547.1
LiTS-FactInference Method=MCTS2026.0547.1
ReActInference Method=Single Trajectory, Temperature=02026.0539.2
Raw SiblingInference Method=Beam Search2026.0539.2
No MemoryInference Method=Beam Search2026.0527.5