List your company

Web Browsing and Navigation on WebWalkerQA

71.7Average AccuracyOpenAI-o3
1.08419.41737.7556.083Jan 29, 2026

Evaluation results

MethodLinks
OpenAI-o32026.0171.7———
Claude-4-Sonnet2026.0161.7———
Reagent-UBackbone=Qwen3-8B2026.0146.249.246.843.3
Reagent-RBackbone=Qwen3-8B2026.0145.347.54642.9
Reagent w/o Agent-RRMBackbone=Qwen3-8B2026.0143.544.44541.3
WebDancerBackbone=Qwen2.5-32B2026.0138.444.346.729.2
WebDancerBackbone=Qwen2.5-7B2026.013640.644.128.2
ARPOBackbone=Qwen3-14B2026.013631.142.931
Reagent-CBackbone=Qwen3-8B, Inference Type=Direct Inference2026.0135.535.638.132.4
Search-o1Backbone=QwQ-32B-Preview2026.0134.143.13527.1
ARPOBackbone=Qwen3-8B2026.0130.526.733.329.6
Qwen3-8BBackbone=Qwen3-8B2026.012931.128.628.2
Atom-SearcherBackbone=Qwen2.5-7B2026.0127.931.723.737
WebThinkerBackbone=Qwen3-8B2026.01136.713.116.9
Tree-GRPOBackbone=Qwen2.5-14B2026.0112.811.115.510.8
DeepSeek-R1-671BBackbone=DeepSeek-R1-671B2026.0110511.811.3
o1-preview2026.019.911.910.47.9
QwQ-32BBackbone=QwQ-32B2026.013.87.52.14.2