Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Preference Prediction on JudgeBench

74.8Positional Consistent Accuracy

MRRG

20.61634.68348.7562.817Apr 15, 2026Apr 28, 2026May 11, 2026May 24, 2026Jun 6, 2026Jun 19, 2026Jul 2, 2026
Updated 23d ago

Evaluation Results

MethodLinks
2026.07
74.8
2026.07
74
2026.07
67.8
2026.07
67.7
2026.07
65.1
2026.07
64
2026.04
63.9
2026.04
63.5
2026.07
63.1
2026.04
60.9
2026.04
60.8
2026.04
60.1
2026.04
59.2
2026.07
56.9
2026.07
56.5
2026.07
55.1
2026.07
54.3
2026.07
52
2026.07
47.4
2026.07
40.9
2026.07
40.5
2026.04
39.8
2026.07
37.1
2026.04
35.8
2026.04
35.2
2026.07
35.1
2026.07
35.1
2026.07
32.3
2026.07
31.1
2026.04
22.7