OpenResearcher: A Fully Open Pipeline for Long-Horizon Deep Research Trajectory Synthesis
🤖 AI Summary
Current deep research agent training is hindered by the scarcity of reproducible, low-cost long-horizon trajectory data and reliance on proprietary APIs, limiting scalability. This work proposes the first fully offline, open-source framework for synthesizing research trajectories: leveraging a 15M-document corpus, it decouples corpus initialization from multi-turn generation and simulates research behaviors using three browser primitives—search, open, and find. High-quality trajectories (97K) are distilled from GPT-OSS-120B for supervised fine-tuning. The approach enables controlled experimentation, revealing the impact of data filtering, agent configuration, and retrieval success rate on answer accuracy. It achieves 54.8% accuracy on BrowseComp-Plus (+34.0 percentage points) and maintains state-of-the-art performance on BrowseComp, GAIA, and xBench-DeepSearch.
Application Category
📝 Abstract
Training deep research agents requires long-horizon trajectories that interleave search, evidence aggregation, and multi-step reasoning. However, existing data collection pipelines typically rely on proprietary web APIs, making large-scale trajectory synthesis costly, unstable, and difficult to reproduce. We present OpenResearcher, a reproducible pipeline that decouples one-time corpus bootstrapping from multi-turn trajectory synthesis and executes the search-and-browse loop entirely offline using three explicit browser primitives: search, open, and find, over a 15M-document corpus. Using GPT-OSS-120B as the teacher model, we synthesize over 97K trajectories, including a substantial long-horizon tail with 100+ tool calls. Supervised fine-tuning a 30B-A3B backbone on these trajectories achieves 54.8\% accuracy on BrowseComp-Plus, a +34.0 point improvement over the base model, while remaining competitive on BrowseComp, GAIA, and xbench-DeepSearch. Because the environment is offline and fully instrumented, it also enables controlled analysis, where our study reveals practical insights into deep research pipeline design, including data filtering strategies, agent configuration choices, and how retrieval success relates to final answer accuracy. We release the pipeline, synthesized trajectories, model checkpoints, and the offline search environment at https://github.com/TIGER-AI-Lab/OpenResearcher.
Problem
Research questions and friction points this paper is trying to address.
long-horizon trajectories
deep research agents
data collection pipeline
reproducibility
offline search environment
Innovation
Methods, ideas, or system contributions that make the work stand out.
long-horizon trajectory synthesis
offline research environment
browser primitives
deep research agent
reproducible pipeline
🔎 Similar Papers
No similar papers found.
