Humanity's Last Exam

Humanity's Last Exam (HLE) is a rigorous multi-disciplinary benchmark of expert-level questions spanning dozens of academic fields, designed to probe the frontier of AI capability. Learn more →

📂

Load the HLE dataset

Drag & drop your .parquet file here, or click anywhere to browse.
Get the dataset from HuggingFace (cais/hle).

🔒 Your file stays on your device. This viewer runs entirely in your browser. The dataset file is never uploaded, transmitted, or sent anywhere — it is only read locally.
⚠️ Please do not publicly share or redistribute the raw dataset file. HLE's value as a benchmark depends on its questions remaining unseen by AI systems. Publicly posting or distributing the file risks it being scraped as training data, permanently undermining the integrity of the benchmark.

This is a low-effort, entirely vibe coded app thrown together by Callum Gare — no dataset content fed to an AI in the making of — not affiliated with CAIS or the HLE project — About HLE