Subject-driven Text-to-Image Generation
Benchmarks
DatasetSOTA methodMetricTrendResultsLast updated
79.5CLIP-T Score
6
Jun 2, 2026
0.782CLIP-T
6
Jun 2, 2026
0.801CLIP-B-I
6
Apr 2, 2026
80.4Subject Alignment Win Rate
6
Feb 26, 2026
0.703DINO Score
5
May 12, 2026
37Subject Fidelity
5
Mar 4, 2026
64Subject Alignment
5
Feb 26, 2026
66.9Preference Rate
4
Apr 13, 2026
77.4CLIP-I
3
Feb 26, 2026
0.859CLIP Score
2
Jun 4, 2026
0.8CLIP Score
2
Jun 4, 2026
33.14CLIP I2T Score
2
Apr 21, 2026
68Subject Fidelity
2
Feb 26, 2026