← Benchmark report cards
Measured benchmarks
FrontierMath
Published by Epoch AI. Source page ↗ · Data we read (file download) ↗
Held-out problems; Epoch AI runs the eval. OpenAI funded the benchmark and has access to much of it.
Facts
| Fact | Value | Where it comes from |
|---|---|---|
| Who runs the evaluation | an independent organisation | curated by aiqbench, reviewed 2026-10-09 |
| Test set exposure | held out | curated by aiqbench, reviewed 2026-10-09 |
| Data licence | CC-BY-4.0 | the board's attribution record |
| Newest published result | Jun 8, 2026 | the ingested rows |
| Last pulled by aiqbench | 21 minutes ago · every 12 h | the sync log |
| Models on this axis | 101 | the ingested rows |
| Rows with a published interval | 98% of rows | the ingested rows |
| Headroom at the top | Best 52.4%, 47.6% below the ceiling; top 10 within 12.1%. | computed from the ingested rows |
| Axes it tracks closely |
| Spearman ρ over shared models (see Clusters) |
Last frontier jumps
- Mar 5, 2026 · gpt-5.4-pro-2026-03-05 · 50.0%
- Dec 11, 2025 · gpt-5.2-2025-12-11 · 40.7%
- Aug 7, 2025 · gpt-5-2025-08-07 · 32.4%
Where it appears
Weighted in the built-in templates: All sources ×1.
View this axis alone in the console →Attribution
Source: Epoch AI, 'AI Benchmarking Hub'. Published online at epoch.ai. Retrieved from https://epoch.ai/benchmarks. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Modified by aiqbench: normalised into aiqbench's common board format; we ingest the 'frontiermath.csv' table and rescale its 0-1 scores to percent. Evaluation run by Epoch AI.