List your company

Web Navigation on WebVoyager 1.0 (test)

62.5AllrecipesMistral 24B distilled student agent
9.04422.92236.850.678Jan 25, 2024May 29, 2024Oct 1, 2024Feb 3, 2025Jun 8, 2025Oct 11, 2025Feb 13, 2026

Evaluation results

MethodLinks
Mistral 24B distilled student agentParameters=24B, Evaluator=GPT-4o2026.0262.563.263.659.562.55547.57076.777.166.760.57568.658.764.5
WilburReference=Lutz et al., 20242026.026043.960.551.22238.659.151.1868103967.453.565.252.6
WebVoyagerText-only*Backbone=GPT-4, Input=Accessibility tree, Evaluator=GPT-4V (full trajectory)2024.0157.843.136.450.463.42.338.624.666.745.27.162.675.23160.244.3
WebVoyager GPT-4o*Backbone=GPT-4o, Evaluator=GPT-4V (full trajectory)2024.0156.353.756.660.557.743.94465.182.254.828.656.963.642.665.255.5
WebVoyagerBackbone=GPT-4o, Reference=He et al., 20242026.0256.353.756.660.557.743.94465.182.254.828.656.963.642.665.255.5
WebVoyagerText-onlyBackbone=GPT-4, Input=Accessibility tree, Evaluator=Human expert2024.0155.631.734.932.6612.336.423.862.845.27.16167.420.958.740.1
WebVoyagerBackbone=GPT-4V, Evaluator=Human expert2024.0153.358.565.151.263.443.238.673.865.161.959.570.776.744.26359.1
WebVoyager*Backbone=GPT-4V, Evaluator=GPT-4V (full trajectory)2024.0151.152.962.85259.332.64757.971.360.351.664.377.555.860.957.1
WebVoyagerBackbone=GPT-4V, Reference=He et al., 20242026.0251.152.962.85259.332.64757.971.360.351.664.377.555.860.957.1
WebVoyager Claude*Backbone=Claude-3-Opus, Evaluator=GPT-4V (full trajectory)2024.0145.958.658.15556.91946.268.271.366.715.155.372.953.551.552.8
WebVoyagerBackbone=Claude, Reference=He et al., 20242026.0245.958.658.15556.91946.268.271.366.715.155.372.953.551.552.8
GPT-4 (All Tools)Backbone=GPT-4, Evaluator=Human expert2024.0111.117.144.21448.822.731.83125.69.52.453.760.537.252.230.8