WebWalker is a multi-agent framework developed by Alibaba Group's Tongyi Laboratory, used to assess the performance of large language models (LLMs) in web browsing tasks. The framework systematically extracts high-quality data by simulating human web browsing behavior through exploration and evaluation paradigms. The primary advantage of WebWalker lies in its innovative web browsing capabilities, which can delve into multi-layered information, addressing the shortcomings of traditional search engines when handling complex queries. This technology is pivotal in enhancing the performance of language models in open-domain question-answering scenarios, especially when multi-step information retrieval is required. The development of WebWalker aims to advance the application and development of language models in the field of information retrieval.