List your company

Multi-turn Refinement on DeepResearch Bench 100 follow-up instances

15.35Instance ScoreSCAFFOLDAGENT
12.54213.2711414.729Jun 18, 2026

Evaluation results

MethodLinks
SCAFFOLDAGENTSetting=SCAFFOLDAGENT2026.0615.3513.414.5515.114.272.6
ReAct-FWSetting=ReAct-FW2026.0613.812.212.756.611.4157.76
ReAct-GRSetting=ReAct-GR2026.0612.6510.8511.5510.913.7559.7