Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

MATH-Vision

Benchmarks

Task NameDataset NameSOTA ResultTrend
Multimodal ReasoningMATH-Vision (full)
Accuracy62.7
38
ReasoningMath Vision
Pass@164.1
32
Mathematical ReasoningMATH-Vision
Accuracy30.4
32
Mathematical problem solvingMATH-Vision (test)
Accuracy68.8
26
GeometryMATH-Vision (mini)
Score40.79
19
ReasoningMath Vision
Pass@475.7
16
Multimodal Mathematical ReasoningMATH-Vision
Accuracy51.2
12
Mathematical Visual Question AnsweringMATH-Vision Full (test)
Relaxed Accuracy73.3
12
Mathematical ReasoningMATH-Vision mini (test)
ALG42.11
8
Multimodal Mathematical ReasoningMATH-Vision (testmini)
Alg Score21.05
8
Showing 10 of 10 rows