iiRecord
Agentic AI Atlas · Humanity's Last Exam (HLE)
benchmark:hlea5c.ai
II.
Benchmark overview

benchmark:hle

Reference · live

Humanity's Last Exam (HLE) overview

Closed-book frontier-knowledge benchmark of expert-authored questions across math, sciences, and humanities, designed to remain unsaturated by current frontier models.

BenchmarkOutgoing · 4Incoming · 0

Attributes

displayName
Humanity's Last Exam (HLE)
homepageUrl
kind
knowledge
targetsKind
ModelVersion
description
Closed-book frontier-knowledge benchmark of expert-authored questions across math, sciences, and humanities, designed to remain unsaturated by current frontier models.

Outgoing edges

applies_to2
covers2

Incoming edges

None.