HLE Diamond
A high-difficulty subset of the HLE (Humanity's Last Exam) benchmark, with questions authored by scholars worldwide and spanning multiple disciplines, designed to test the upper limit of a model's reasoning on expert-level knowledge.
Timeline (last 90 days)
The Comprehensive Intelligence leaderboard includes nine highly difficult tests, and Carbon took first place in four individual categories: highly difficult complex reasoning (HLE Diamond), the scientific terminal evaluation, an all-subject encyclopedic knowledge base, and an automated workflow benchmark.
HLE Diamond is a highly difficult subset of the Humanity's Last Exam benchmark, with questions written by scholars worldwide, covering dozens of disciplines and using stricter scoring standards.
All Facts (2)
The Comprehensive Intelligence leaderboard includes nine highly difficult tests, and Carbon took first place in four individual categories: highly difficult complex reasoning (HLE Diamond), the scientific terminal evaluation, an all-subject encyclopedic knowledge base, and an automated workflow benchmark.
50%UnverifiedHLE Diamond is a highly difficult subset of the Humanity's Last Exam benchmark, with questions written by scholars worldwide, covering dozens of disciplines and using stricter scoring standards.
50%