Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Physics-aware Action Generation on Do-Undo Forward
Loading...
8.41
Instruction Following (IF)
Qwen (20B)
7.4532
7.7016
7.95
8.1984
Dec 15, 2025
Instruction Following (IF)
Identity Preservation (IDP)
Object Consistency (OC)
Updated 1mo ago
Evaluation Results
Method
Method
Links
Instruction Following (IF)
Identity Preservation (IDP)
Object Consistency (OC)
Qwen (20B)
type=Und&Gen
2025.12
8.41
8.81
8.93
Gemini
type=Und&Gen
2025.12
8.34
9.3
9.43
Bagel-think (7B)
type=Und&Gen
2025.12
7.83
8.78
8.99
Do-Undo
type=Und&Gen
2025.12
7.81
8.53
8.71
FluxKontext (7B)
type=Gen
2025.12
7.49
7.51
7.73
Feedback
Search any
task
Search any
task