Math Reasoning
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
34.8Exact-match Accuracy (Ngen=128)
6
Feb 26, 2026
75Exact Match Accuracy (Ngen=128)
6
Feb 26, 2026
47.17Average Score
5
Mar 4, 2026
68.45Accuracy@1
5
Mar 4, 2026
29.93Accuracy@1
5
Mar 4, 2026
44.72Accuracy@1
5
Mar 4, 2026
0.75False Positive Rate (FPR)
5
Feb 26, 2026
1.12FPR
5
Feb 26, 2026
0.62False Positive Rate
5
Feb 26, 2026
3.75FPR
5
Feb 26, 2026
36.87AIME 2025 Avg Score @16
5
Feb 26, 2026
37.91avg@16
5
Feb 26, 2026
0.613AUC-ROC
5
Feb 26, 2026
51.24Accuracy @32
4
Jul 1, 2026
0.113Accuracy (avg@32)
4
Jul 1, 2026
20.83Gap
4
Jun 11, 2026
24Gap
4
Jun 11, 2026
8,732.8Average Length
4
Jun 4, 2026
802.3Average Output Length
4
Jun 4, 2026
40.6Best Answer Rate @20
4
May 29, 2026
40AIME 2024 Score
4
May 28, 2026
90.2MATH 500 Score
4
May 28, 2026
47.39Pass@1
4
May 26, 2026
22.29Pass@1
4
May 26, 2026
35.72Pass@1
4
May 26, 2026