Print-friendly view · use your browser's Save as PDF option (Cmd/Ctrl-P) to attach this article to a brief.
Humanity's Last Exam
HLE · knowledge benchmark · 2025
Source: https://policywindow.org/wiki/humanitys-last-exam
Generated 2026-05-30T22:12:40 UTC
Summary
3,000+ frontier-difficulty expert-curated questions across all academic disciplines. Designed to remain unsaturated through 2026+.
At a glance
- Score range
- 0–100 % accuracy
- Contamination risk
- low
- Methodology URL
- https://lastexam.ai/
- Saturation status
- active
Details
Center for AI Safety + Scale AI collaboration. Frontier models 8-22% at launch. Replaces MMLU as the de-facto knowledge ceiling.
How to cite this article
APA
Policy Window. (2025). Humanity's Last Exam [Wiki article — Benchmark]. https://policywindow.org/wiki/humanitys-last-exam
Chicago
Policy Window. 2025. "Humanity's Last Exam." Wiki article (Benchmark). https://policywindow.org/wiki/humanitys-last-exam.
Harvard
Policy Window (2025) 'Humanity's Last Exam', Wiki article — Benchmark, available at: https://policywindow.org/wiki/humanitys-last-exam.
OSCOLA
Policy Window, 'Humanity's Last Exam' (Wiki article — Benchmark, 2025) <https://policywindow.org/wiki/humanitys-last-exam> accessed [date].
BibTeX
@misc{policywindow-humanitys-last-exam,
title = {Humanity's Last Exam},
author = {Policy Window},
year = {2025},
howpublished = {HLE (2025)},
url = {https://policywindow.org/wiki/humanitys-last-exam},
note = {Primary source: https://lastexam.ai/}
}