Home › Benchmarks › FrontierMath
FrontierMath
Benchmark
Capability
Research-grade mathematics — built unsolvable, already yielding.
You're looking at a benchmark record: a test the industry uses to measure chips or models — and what it actually tells you.
Epoch AI commissioned original problems from working mathematicians (Fields medalists among the reviewers) that take experts hours or days. Launch scores were ~2%; frontier systems passed 25–40% on the easier tiers within a year — while a disclosure that OpenAI funded the set sharpened norms about benchmark independence.
How this connects to the rest of the stack
This record's links at a glance — click a group to jump to it.
Related
Sources
Record reviewed as of 2026-01-15 — see Methodology.