Benchmarks
Benchmarks

FrontierMath (Tiers 1-3)

Mathunit % · normalized over [0, 94]

Novel unpublished research-level math problems, easier tiers.

#ModelSourceScoreNormalized
Score distribution
14 tracked results across the normalization window
094
Score vs. parameters
Open-weights models, log-x params
No open-weights models with disclosed parameter counts have a score here yet.