OpenSeeker: Democratizing Frontier Search Agents by Fully Open-Sourcing Training Data

Dimensions overall score 9.0

• OpenSeeker democratizes frontier search agent research by providing fully open-source training data and model weights.market 95%
• OpenSeeker is the first fully open-source search agent (model and data) that achieves frontier-level performance.technical 95%
• OpenSeeker achieves state-of-the-art performance across multiple benchmarks including BrowseComp, BrowseComp-ZH, xbench-DeepSearch, and WideSearch.result 90%
• OpenSeeker significantly outperforms the second-best fully open-source agent DeepDive (29.5% vs. 15.3% on BrowseComp).result 95%
• OpenSeeker surpasses industrial competitor Tongyi DeepResearch on BrowseComp-ZH (48.4% vs. 46.7%).result 90%
• OpenSeeker achieves this performance with only 11.7k synthesized training samples in a single training run.method 95%
• The model uses fact-grounded scalable controllable QA synthesis that reverse-engineers the web graph via topological expansion and entity obfuscation.method 90%
• The model's performance heavily depends on the quality of web data, and potential biases in dataset creation may impact results.limitation 85%
• OpenSeeker is the first fully open-source search agent (model and data) that achieves frontier-level performance.technical 95%
• OpenSeeker utilizes fact-grounded scalable controllable QA synthesis to generate training data.method 90%
• OpenSeeker employs denoised trajectory synthesis with a retrospective summarization mechanism to improve action generation.method 90%
• OpenSeeker achieves state-of-the-art performance across multiple benchmarks including BrowseComp, BrowseComp-ZH, xbench-DeepSearch, and WideSearch.result 95%

Competitive landscape

Competitor map is still being generated for this paper. Enable generation or check back soon.

Startup potential card