Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Compositional Image Generation on DrawBench Held-out (test)
Loading...
97
GenEval
RAM
62.68
71.59
80.5
89.41
May 11, 2026
GenEval
Aesthetic Score
DeQA
ImageReward
HPSv2
PickScore (Image Quality)
Updated 22d ago
Evaluation Results
Method
Method
Links
GenEval
Aesthetic Score
DeQA
ImageReward
HPSv2
PickScore (Image Quality)
RAM
# Steps=270
2026.05
97
5.38
4.09
1.19
29
22.52
Flow-GRPO
# Steps=> 5k
2026.05
95
5.25
4.01
1.03
27
22.37
DiffusionNFT†
# Steps=900
2026.05
95
4.98
4.1
0.3
24
21.59
AWM†
# Steps=300
2026.05
83
5.14
3.75
0.67
24
22.04
SD3.5M
2026.05
64
5.41
4.08
0.82
28
22.4
Feedback
Search any
task
Search any
task