Human Preference Evaluation
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
73.33Win Rate
2
May 20, 2026
73.33Win Rate
2
May 20, 2026
48Good Count
2
Mar 27, 2026
82Preference
2
Mar 13, 2026
48.8Preference Share: Significantly Better
2
Feb 26, 2026
59Preference Rate
2
Feb 26, 2026
71Preference Rate
2
Feb 26, 2026
64Win Rate (Ours Better)
1
Jun 16, 2026
0.215Preference Rate (≫)
1
Feb 26, 2026