Model results

FrontierMath

Choose a shared evaluation protocol to compare model scores. Versions, sources and configurations stay separate.

Atlas snapshot captured 10/19/2026, 10:20:28 AM UTC. This is an import timestamp, not the evaluation date. Coverage may be incomplete.

Reviewed recommendation and limits

Use the relevant tier and private/public split for difficult mathematical reasoning.

Each tier, release and public/private split separately. Imported groups below may use different protocols; this review does not certify each run. Difficulty is specialized; tier, dataset release, tools and grader must match before comparing scores.

Publisher results Epoch benchmarking methods

← All evaluation groups

FrontierMath Tiers 1–3 v2 · private

Source: Epoch AI Benchmarking Hub
Version: tiers-1-3-v2-private
Measurement date: Not reported
Metric: % · Higher is better
Evaluation scope: Epoch AI run · task v2 · effort 32K · result date not supplied

Exact score ties share a position. Positions do not establish statistical significance. Missing metadata is shown as unreported; matching imported metadata is not independent verification of the protocol.

#1Claude Opus 4.5
anthropic/claude-opus-4-5
34.4%
View modelEvidence independent
#2Claude Sonnet 4.5
anthropic/claude-sonnet-4-5
23.9%
View modelEvidence independent
#3Claude Opus 4.1
anthropic/claude-opus-4-1
12.6%
View modelEvidence independent