Math Reasoning
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
Math Reasoning AIME24, AIME25, HMMT25
78.4AIME24 Score
30
Math Reasoning Overall
63.8Mean@16
24
Math Reasoning
87.6Accuracy
20
Math Reasoning Long Q, Long A (test)
0.65Pass@1
15
Math Reasoning Out-domain (SVAMP, Mathematics, SimulEq) (test)
79.6SVAMP Accuracy
8
Math Reasoning In-domain (GSM8K, MATH, NumGLUE) (test)
69.1GSM8K Accuracy
8
Overall Average Math Reasoning
54.54Pass@1
6
Math Reasoning Aggregate
40.08Avg@32
6
Math-Reasoning (test)
20.83Gap
4
Math-Reasoning (dev)
24Gap
4
Math Reasoning (test)
85.4Classification Accuracy
4
Math Reasoning 1.5B model (val)
69.4Validation Accuracy
3
Math-Reasoning I
96.62Accuracy
1