AIME
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
AIME-25
55.3Pass@4
25
AIME 24
60Pass@4
25
AIME 25
86.4Accuracy (AIME 25)
25
AIME25
84.17Accuracy
25
AIME25
70.42Accuracy
25
AIME 24, 25
76Accuracy
25
AIME 24
14Accuracy
25
AIME 2024
93.33Accuracy
25
AIME24
6.16Speedup Ratio
24
AIME25
63.3Accuracy
24
AIME 2025
59.2Accuracy
24
AIME 2025 (held-out)
59.07Pass@1
24
AIME 2024
73.3Accuracy
24
AIME 24
8.33Avg@16
24
AIME 2024
66.67Accuracy (%)
24
AIME24
73.3Accuracy
24
AIME 25
6.99Speedup
24
AIME 2025
35.6Mean Score (AIME 2025)
24
AIME 2024
42Mean@16
24
AIME25
37.4Mean@16 Score
24
AIME 2025 1
25Accuracy
24
AIME 2024 1
30Accuracy
24
AIME combined 2020-2025
0.368R2
24
AIME 24
71.7Avg@16 Accuracy
24
AIME 2025
80Accuracy
24