Web agents require massive trajectories to generalize, yet real-world training is constrained by network latency, rate limits, and safety risks. We introduce \textbf{WebWorld} series, the first open-web simulator trained at scale. While existing simulators are restricted to closed environments with thousands of trajectories, WebWorld leverages a scalable data pipeline to train on 1M+ open-web interactions, supporting reasoning, multi-format data, and long-horizon simulations of 30+ steps. For intrinsic evaluation, we introduce WebWorld-Bench with dual metrics spanning nine dimensions, where WebWorld achieves simulation performance comparable to Gemini-3-Pro. For extrinsic evaluation, Qwen3-14B trained on WebWorld-synthesized trajectories improves by +9.2\% on WebArena, reaching performance comparable to GPT-4o. WebWorld enables effective inference-time search, outperforming GPT-5 as a world model. Beyond web simulation, WebWorld exhibits cross-domain generalization to code, GUI, and game environments, providing a replicable recipe for world model construction.

翻译：网络智能体需要海量轨迹以实现泛化，然而现实世界的训练受到网络延迟、速率限制和安全风险的制约。我们推出\textbf{WebWorld}系列，这是首个大规模训练的开源网络模拟器。现有模拟器局限于封闭环境，仅能处理数千条轨迹，而WebWorld利用可扩展的数据流水线，在超过100万次的开放网络交互上进行训练，支持推理、多格式数据以及超过30步的长时程模拟。在内在评估方面，我们引入WebWorld-Bench，其包含覆盖九个维度的双重指标，WebWorld在该基准上达到了与Gemini-3-Pro相当的模拟性能。在外在评估方面，基于WebWorld合成轨迹训练的Qwen3-14B模型在WebArena上的性能提升了+9.2\%，达到与GPT-4o相当的水平。WebWorld支持高效的推理时搜索，作为世界模型其性能超越了GPT-5。除了网络模拟，WebWorld还展现出对代码、图形用户界面和游戏环境的跨领域泛化能力，为世界模型的构建提供了一个可复现的方案。