Epoch AI releases a revised FrontierMath

Released a revised FrontierMath — in Tier 4, 12 problems revised and 7 removed (in Tiers 1–3, 123 problems revised and 5 removed)

Why it matters

Large-scale errors were found even in a top-difficulty benchmark, raising doubts about the reliability of its review once again

Source — Primary sources first; secondary reports are labelled “Press”

Officialepoch.aihttps://epoch.ai/benchmarks/frontiermath-tier-4-v2

Added in Week 35, 2026 · See that week

Same week

Third Circuit hears oral argument in Thomson Reuters v. RossPressCohere releases North Mini CodeOfficialLiu et al. release ABC-BenchPaperElastic publishes vector search benchmarksOfficial

Nearby on the same axis

Liu et al. release ABC-BenchPaperBioSecBench-Refusal releasedPaperLaude Institute releases Terminal-Bench 2.1Officialimo-2026 project runs 7 frontier models on IMO 2026, publishes audit trailsOfficial