Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Text-to-Image Generation on GenEval (Object Counting Compositionality)
Loading...
92
Accuracy (2 objects)
OSI
80.56
83.53
86.5
89.47
May 14, 2026
Accuracy (2 objects)
Accuracy (3 objects)
Accuracy (4 objects)
Accuracy (5 objects)
Accuracy (6 objects)
Average Compositional Accuracy
Updated 19d ago
Evaluation Results
Method
Method
Links
Accuracy (2 objects)
Accuracy (3 objects)
Accuracy (4 objects)
Accuracy (5 objects)
Accuracy (6 objects)
Average Compositional Accuracy
OSI
Backbone=FLUX
2026.05
92
71
64
40
40
61
TACA
Backbone=FLUX
2026.05
89
68
56
31
20
53
OSI
Backbone=SD3.5
2026.05
89
80
68
46
35
64
PLADIS
Backbone=FLUX
2026.05
87
71
56
31
20
53
TACA
Backbone=SD3.5
2026.05
87
74
55
47
26
58
Base
Backbone=SD3.5
2026.05
82
72
59
38
24
55
Base
Backbone=FLUX
2026.05
81
63
44
29
18
47
Feedback
Search any
task
Search any
task
