AIME
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
AIME 2026
59.1Average Success Rate (avg@32)
29
AIME25
72.2Pass@1
29
AIME 2025
86.67Accuracy
29
AIME 2025
17.07Latency (s)
29
AIME 2024
29.58Latency (s)
29
AIME 2023
32.93Latency (s)
29
AIME 2025
29,494.1Input Tokens
29
AIME 2024
25,573.15Input Tokens
29
AIME 2023
32,881.13Input Tokens
29
AIME 2024
93.4Pass@1
29
AIME25
40Pass@8
29
AIME 25
69.7Mean@12 Score
28
AIME24
77.5Mean@12 Score
28
AIME
28.7Accuracy
28
AIME 26
70Mean@8
28
AIME 25
57Exact Match
28
AIME24
70Exact Match
28
AIME
85Accuracy
28
AIME25
94.6AIME25 Accuracy
28
AIME 2024
100Final Accuracy
28
AIME25
54.7Accuracy
27
AIME26
11,381Thinking Tokens
27
AIME 24/25/26 (val)
54.79Avg@16 Accuracy
27
AIME 25
83.3Accuracy
27
AIME 24
90Accuracy
27