Reward Modeling
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
96.7Safety Score
284
Jul 7, 2026
97.8Accuracy
166
Apr 10, 2026
96Accuracy
137
Jun 9, 2026
89.3Accuracy
120
Apr 10, 2026
93.3Accuracy
117
Jun 1, 2026
89.89Precise IF Score
90
Jun 9, 2026
0.978Average Score
89
Mar 4, 2026
90.3Accuracy
82
Feb 26, 2026
92.1Accuracy
72
Mar 4, 2026
84.11Chat Score
72
Jun 8, 2026
79.8Accuracy
72
Jun 1, 2026
57.5Accuracy
70
Feb 26, 2026
86.5Average Score
67
Apr 28, 2026
0.077MAE
65
May 8, 2026
83.15Accuracy
62
May 29, 2026
39.7Accuracy
52
Feb 26, 2026
35.7Accuracy
52
Feb 26, 2026
0.154EF
52
Feb 26, 2026
0.305EF
52
Feb 26, 2026
74.56Overall Accuracy
45
Feb 26, 2026
71.2Accuracy
45
Jun 1, 2026
78.5Accuracy
42
May 29, 2026
96.4Accuracy
42
May 29, 2026
99.2Chat Score
41
May 19, 2026
87.2Correlation
40
Mar 4, 2026