DeepMath
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
DeepMath
66.2Accuracy
58
DeepMath
70.5Pass@1
44
DeepMath 2025 (test)
55.8Pass@1
32
DeepMath
78Peak Accuracy
24
DeepMath hard (held-out)
73Pass@1
23
DeepMath-103K (test)
43.27Accuracy
16
DEEPMATH 128 samples
35.93Top-1 Accuracy
12
DeepMath500
69Pass@1 Rate
12
DeepMath (test)
62Pass@1
12
DeepMath (test)
79.5Accuracy
10
DEEPMATH-103K
95.5Accuracy
8
DeepMath
94FR (Fetch Rate)
8
DeepMath 103K subset
65.9Accuracy
6
DeepMATH (test)
83.4MATH 500 Score
4