Frontier Leaderboards
Legacy Leaderboards
2025 Scale AI. All rights reserved.
Humanity's Last Exam (Text Only)
Models evaluated on text-only HLE questions
Last updated: April 16, 2025
Performance Comparison
1
20.57±1.71Calib Err: 36
1
19.78±1.68Calib Err: 37
1
18.90±1.65Calib Err: 58
1
18.38±1.64Calib Err: 71
1
18.38±1.64Calib Err: 70
6
14.53±1.49Calib Err: 58
6
13.37±1.44Calib Err: 80
6
12.58±1.40Calib Err: 81
6
11.75±1.36Calib Err: 74
8
10.31±1.28Calib Err: 81
10
DeepSeek R1
8.54±1.18Calib Err: 73
10
7.89±1.14Calib Err: 81
11
7.75±1.13Calib Err: 84
11
7.71±1.13Calib Err: 82
11
6.55±1.04Calib Err: 82
12
5.80±0.99Calib Err: 83
15
5.34±0.95Calib Err: 84
15
4.97±0.92Calib Err: 88
16
4.55±0.88Calib Err: 80
16
4.55±0.88Calib Err: 87
16
4.41±0.87Calib Err: 84
16
4.36±0.86Calib Err: 78
16
4.32±0.86Calib Err: 83
16
4.32±0.86Calib Err: 80
17
3.76±0.80Calib Err: 83
25
2.32±0.64Calib Err: 88