🌐 WebAgent for Information Seeking bulit by Tongyi Lab: WebWalker & WebDancer & WebSailor https://arxiv.org/pdf/2507.02592

🤗 WebSailor ｜ 🤗 WebDancer-QwQ-32B | ModelScope WebDancer-QwQ-32B | 🤗 WebWalkerQA

You can check the paper of WebDancer and WebWalker and WebSailor.

💥 💥 💥 Stay tuned for more updates! We are working on building native agentic model based on the Browser and more open-domain environments!

• WebSailor (Preprint 2025) - WebSailor: Navigating Super-human Reasoning for Web Agent
• WebDancer (Preprint 2025) - WebDancer: Towards Autonomous Information Seeking Agency
• WebWalker (ACL 2025) - WebWalker: Benchmarking LLMs in Web Traversal

• 2025.07.03 🔥🔥🔥We release WebSailor, an agentic search model specialized in performing extremely complex information seeking tasks, achieving open-source SOTA on some of the most difficult browsing benchmarks. WebSailor topped the HuggingFace daily papers.
• 2025.06.23 🔥🔥🔥The model, interactive demo, and some of the data of WebDancer have been open-sourced. You're welcome to try them out!
• 2025.05.29 🔥🔥🔥We release WebDancer, a native agentic search model towards autonomous information seeking agency and Deep Research-like model.
• 2025.05.15 WebWalker is accepted by ACL 2025 main conference.
• 2025.01.14 We release WebWalker, a benchmark for LLMs in web traversal and a multi-agent framework for information seeking.

• A complete post-training methodology enabling models to engage in extended thinking and information seeking, ultimately allowing them to successfully complete extremely complex tasks previously considered unsolvable.
• Introduces SailorFog-QA, a scalable QA benchmark with high uncertainty and difficulty, curated with a novel data synthesis method through graph sampling and information obfuscation. Example SailorFog-QA data samples can be found at: WebSailor/dataset/sailorfog-QA.jsonl
• Effective post-training pipeline consisting of (1) high-quality reconstruction of concise reasoning from expert trajectories for clean supervision, (2) a two-stage training process involving an RFT cold start stage, followed by Duplicating Sampling Policy Optimization (DUPO), an efficient agentic RL algorithm excelling in effectiveness and efficiency.
• WebSailor-72B significantly outperforms all open-source agents and frameworks while closing the performance gap with leading proprietary systems, achieving a score of 12.0% on BrowseComp-en, 30.1% on BrowseComp-zh, and 55.4% on GAIA.
• The checkpoint is coming soon.

• Native agentic search reasoning model using ReAct framework towards autonomous information seeking agency and Deep Research-like model.
• We introduce a four-stage training paradigm comprising browsing data construction, trajectory sampling, supervised fine-tuning for effective cold start, and reinforcement learning for improved generalization, enabling the agent to autonomously acquire autonomous search and reasoning skills.
• Our data-centric approach integrates trajectory-level supervision fine-tuning and reinforcement learning (DAPO) to develop a scalable pipeline for training agentic systems via SFT or RL.
• WebDancer achieves a Pass@3 score of 64.1% on GAIA and 62.0% on WebWalkerQA.

You need to enter the WebDancer folder for the following commands.

conda create -n webdancer python=3.12
pip install -r requirements.txt

Download the WebDancer model from 🤗 HuggingFace and deploy it using the provided scripts with sglang.

cd scripts
bash deploy_model.sh WebDancer_PATH

Note: Replace WebDancer_PATH with the actual path to the downloaded model.

Edit the following keys in WebDancer/scripts/run_demo.sh:

• GOOGLE_SEARCH_KEY, you can get it from serpapi or serper.
• JINA_API_KEY, you can get it from jina.
• DASHSCOPE_API_KEY, you can get it from dashscope.

Then, launch the demo with Gradio to interact with the WebDancer model:

cd scripts
bash run_demo.sh

We provide demos for BrowseComp-en, BrowseComp-zh and Daily Use. Our model can complete highly difficult and uncertain tasks requiring massive information acquisition and complex reasoning.

We provide demos for WebWalkerQA, GAIA and Daily Use. Our model can execute the long-horizon tasks with multiple steps and complex reasoning, such as web traversal, information seeking and question answering.

The content of this project itself is licensed under LICENSE.

If this work is helpful, please kindly cite as:

@misc{li2025websailor,
      title={WebSailor: Navigating Super-human Reasoning for Web Agent},
      author={Kuan Li and Zhongwang Zhang and Huifeng Yin and Liwen Zhang and Litu Ou and Jialong Wu and Wenbiao Yin and Baixuan Li and Zhengwei Tao and Xinyu Wang and Weizhou Shen and Junkai Zhang and Dingchu Zhang and Xixi Wu and Yong Jiang and Ming Yan and Pengjun Xie and Fei Huang and Jingren Zhou},
      year={2025},
      eprint={2507.02592},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2507.02592},
}
@misc{wu2025webdancer,
      title={WebDancer: Towards Autonomous Information Seeking Agency},
      author={Jialong Wu and Baixuan Li and Runnan Fang and Wenbiao Yin and Liwen Zhang and Zhengwei Tao and Dingchu Zhang and Zekun Xi and Yong Jiang and Pengjun Xie and Fei Huang and Jingren Zhou},
      year={2025},
      eprint={2505.22648},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2505.22648},
}
@misc{wu2025webwalker,
      title={WebWalker: Benchmarking LLMs in Web Traversal},
      author={Jialong Wu and Wenbiao Yin and Yong Jiang and Zhenglin Wang and Zekun Xi and Runnan Fang and Deyu Zhou and Pengjun Xie and Fei Huang},
      year={2025},
      eprint={2501.07572},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2501.07572},
}

The repo is contributed by Jialong Wu. If you have any questions, please feel free to contact via wujialongml@gmail.com or create an issue.

🔥🔥🔥 We are hiring! Research intern positions are open (based in Hangzhou、Beijing、Shanghai)

📚 Research Area：Web Agent, Search Agent, Agent RL, MultiAgent RL, Agentic RAG

☎️ Contact：yongjiang.jy@alibaba-inc.com