Reproduce “Deep Search in Real-World Web Environments” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Xiaomin Yu, Hongliang Chen, Zhe Huang, Shilin Yan, Shao-Rong Su, Xuqiao Ma, Feiyang Wang, Jiayu Chen and 17 more
OpenDS-World benchmarks agents’ ability to find, access, and connect evidence across live-Web sources, formats, and interactive operations. Annotators build multi-step evidence paths from public videos, then turn them into 200 tasks across eight domains. Under a shared tool setup and budget, GPT-6 Astra answered 78 tasks correctly (39.0%), showing that tested models still struggle on this benchmark.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.