Reproduce “FrontierPhysics: Evaluating Agents for End-to-End Physics Research” 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
Bingran You, Wenjun Ke, Ria Rosenauer, Bo Liu, Wenbo Chen, Min Chen, Shuqi Xu, Yue Ma and 64 more
FrontierPhysics evaluates whether AI agents can conduct open-ended physics research from high-level briefs. The benchmark uses physicist-contributed research tasks and grades structured outputs with deterministic checks, alongside manuscripts and research trajectories against expert-authored rubrics. Passing requires all verifier tests and blocking criteria. The top tested configuration passed 46/168 trials (27.4%), measuring performance under this protocol rather than human publication acceptance.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.