Reproduce “Humanity's Last Exam” 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
Long Phan, Alice Gatti, Ziwen Han, Nathaniel Li, Josephina Hu, Hugh Zhang, Chen Bo Calvin Zhang, Mohamed Shaaban and 1171 more
HLE provides a broad benchmark of difficult academic questions designed by subject experts to evaluate models beyond familiar tests. The authors screen questions for model difficulty and use human review, with short answers that can be checked. Gemini 2.0 Flash Thinking scored 6.6% accuracy and 82% RMS calibration error; these results describe benchmark performance, not human equivalence or general intelligence.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.