LuminaBench
Mathematics benchmark / Epoch AI

FrontierMath Open Problems

Lumina VerifiedLumina checked the published sources, versions, model labels, score extraction and conversion. External evaluations were run by their publishers; Lumina has not independently rerun or reproduced them.

Solve rate

Version Erdos-68-Lean-open-problems-v1Verified 6 November 202614 published resultsOfficial source

FrontierMath Open Problems

LuminaBench
Show
14 of 14 models · best variant per model
03006009001,2001,165GPT-6.1 SolMax1,090GPT-6 AstraMax1,090Claude Opus 5.5Max1,088Claude Sonnet 5.5Max1,016GPT-6 SolMax1,016Gemini 4 ArgonHigh942Claude Fable 5Max942Claude Fable 5.1Max942Gemini 3.8 FlashHigh942GPT-5.5xHigh942Grok 4.7xHigh942Kimi K3Max942Claude Opus 5Max942GPT-5.6 SolMax

No comparable per-task resource measurements were published. Missing metrics are omitted from their graph; quality scores remain available.

Results table

Method & coverage

68 Erdős open conjectures; text mathematical statements, Lean proof construction and isolated proof verification. Do not mix independently reported outside-Epoch proof announcements into the benchmark runs.

No comparable per-task resource measurements were published.

How the Lumina conversion works

1,000 + 75 × ((original score − 1.152436974789916) / 1.4856334059128964). The result is rounded once. Scores may tie.

Frozen initial observed text-only cohort; one mean per exact publisher release/API, balancing repeated settings/routes within that release. Original score and protocol remain separate. No inferred variants or future refitting.

Calibration
frontiermath-open-problems-z75-v1 · experimental
Reference mean
1.152436974789916
Population SD
1.4856334059128964
Coverage
Standalone benchmark.
Identity, exclusions & attribution

68 Erdős open conjectures; text mathematical statements, Lean proof construction and isolated proof verification. Do not mix independently reported outside-Epoch proof announcements into the benchmark runs.

Each run ID is retained. Repeated observations are not additional model releases or native reasoning settings. Standard errors are retained only where their mean-score association matches the selected primary metric.

Text mathematical statements/proof/code tools; no image, video, speech or screenshot input protocol.

Standalone text benchmark; no new combined index. Source identity/effort/protocol qualifications remain in Data and CSV.

All published reasoning settings, including Claude Max, are included by default. Display labels use Low and xHigh; source labels remain in the retained records.

Official source · Publication terms / evidence

Publisher terms. Scores and resource figures are attributed to the original publisher.