WebVoyager

WebVoyager is a benchmark dataset used to evaluate the performance of computer use agents (CUAs) in navigating and interacting with web-based graphical user interfaces.

Papers using WebVoyager (4)

Orchard: An Open-Source Agentic Modeling Framework2026DynaWeb: Model-Based Reinforcement Learning of Web Agents2026WebSTAR: Scalable Data Synthesis for Computer Use Agents with Step-Level Filtering2025Thinking vs. Doing: Agents that Reason by Scaling Test-Time Interaction2025