Paper 2603.15594
OpenSeeker: Democratizing Frontier Search Agents by Fully Open-Sourcing Training Data
- Published
- Mar 2026
- Research lab
- Independent
- Citations
- 21
- GitHub
- 766 stars
01 In brief
Summary
OpenSeeker is the first fully open-source search agent (model and data) achieving frontier-level performance, developed by an academic team at Shanghai Jiao Tong University.
It uses two core innovations: fact-grounded scalable controllable QA synthesis, which reverse-engineers the web graph via topological expansion and entity obfuscation to generate complex multi-hop reasoning tasks, and denoised trajectory synthesis, which uses retrospective summarization to generate high-quality actions.
Trained with simple SFT on only 11.7k synthesized samples, OpenSeeker achieves state-of-the-art results on BrowseComp (29.5%), BrowseComp-ZH (48.4%), xbench-DeepSearch (74.0%), and WideSearch (59.4% item F1).
It outperforms the second-best fully open-source agent DeepDive (29.5% vs 15.3% on BrowseComp) and surpasses industrial competitor Tongyi DeepResearch on BrowseComp-ZH (48.4% vs 46.7%).
The complete training dataset and model weights are open-sourced to democratize search agent research.
The model is initialized from Qwen3-30B-A3B-Thinking-2507, with a 256k context window and a maximum of 200 tool calls.
The dataset includes 10.3k English and 1.4k Chinese samples, with Chinese data averaging 46.35 tool calls per trajectory, exceeding BrowseComp-ZH difficulty.
OpenSeeker demonstrates that strategic data synthesis can bridge the performance gap with industrial-scale efforts, breaking the corporate data monopoly in search agent development.
02 From the paper
Abstract
Deep search capabilities have become an indispensable competency for frontier Large Language Model (LLM) agents, yet the development of high-performance search agents remains dominated by industrial giants due to a lack of transparent, high-quality training data. This persistent data scarcity has fundamentally hindered the progress of the broader research community in developing and innovating within this domain. To bridge this gap, we introduce OpenSeeker, the first fully open-source search agent (i.e., model and data) that achieves frontier-level performance through two core technical innovations: (1) Fact-grounded scalable controllable QA synthesis, which reverse-engineers the web graph via topological expansion and entity obfuscation to generate complex, multi-hop reasoning tasks with controllable coverage and complexity. (2) Denoised trajectory synthesis, which employs a retrospective summarization mechanism to denoise the trajectory, therefore promoting the teacher LLMs to generate high-quality actions. Experimental results demonstrate that OpenSeeker, trained (a single training run) on only 11.7k synthesized samples, achieves state-of-the-art performance across multiple benchmarks including BrowseComp, BrowseComp-ZH, xbench-DeepSearch, and WideSearch. Notably, trained with simple SFT, OpenSeeker significantly outperforms the second-best fully open-source agent DeepDive (e.g., 29.5% v.s. 15.3% on BrowseComp), and even surpasses industrial competitors such as Tongyi DeepResearch (trained via extensive continual pre-training, SFT, and RL) on BrowseComp-ZH (48.4% v.s. 46.7%). We fully open-source the complete training dataset and the model weights to democratize frontier search agent research and foster a more transparent, collaborative ecosystem.