List your company

Web Search on xbench

66Average ScoreOpenAI-o3
6.7222.1137.552.89Jan 29, 2026

Evaluation results

MethodLinks
OpenAI-o32026.0166
Claude-4-Sonnet2026.0164
Reagent-UBackbone=Qwen3-8B2026.0143
Reagent-RBackbone=Qwen3-8B2026.0141
Search-o1Backbone=QwQ-32B-Preview2026.0140
WebDancerBackbone=Qwen2.5-32B2026.0138
DeepSeek-R1-671BBackbone=DeepSeek-R1-671B2026.0132
ARPOBackbone=Qwen3-14B2026.0132
Reagent w/o Agent-RRMBackbone=Qwen3-8B2026.0132
ARPOBackbone=Qwen3-8B2026.0125
Atom-SearcherBackbone=Qwen2.5-7B2026.0121
Reagent-CBackbone=Qwen3-8B, Inference Type=Direct Inference2026.0115
WebThinkerBackbone=Qwen3-8B2026.0113
QwQ-32BBackbone=QwQ-32B2026.0110
Qwen3-8BBackbone=Qwen3-8B2026.019