Inference
Dan Hendrycks et al.

MATH-500 Problem Set

A curated subset of 500 problems from the MATH dataset, covering algebra, counting and probability, geometry, intermediate algebra, number theory, prealgebra, and precalculus.

ModelsModels with a result.7
Top resultBest result on this test.98.8%Ternary Bonsai 2 27B
Top-3 spreadPoints from first to third.4.2 pts
YearYear of release.2021

Results

7 results
#ModelResultThe score from the source.
01Ternary Bonsai 2 27B98.8%
02LongCat-Flash-Lite-Sparse95.8%
03MiniCPM5-2B94.6%
04MiniCPM5-1B91.6%
05LFM2.5-8B-A1B88.8%
06Kanana-2 1.3B Instruct61.4%
07Kanana-2 3B Instruct61.2%