Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Perception on Visual Probe (Pass@4)
Loading...
67.9
Pass@4
AXPO
12.988
27.244
41.5
55.756
May 27, 2026
Pass@4
Updated 6d ago
Evaluation Results
Method
Method
Links
Pass@4
AXPO
Model Size=8B, Backbon...
2026.05
67.9
AXPO
Model Size=4B, Backbon...
2026.05
65.1
Base
Model Size=32B, Backbo...
2026.05
64.2
SFT + GRPO
Model Size=8B, Backbon...
2026.05
63.2
GRPO
Model Size=8B, Backbon...
2026.05
62.3
SFT + GRPO
Model Size=2B, Backbon...
2026.05
60.4
AXPO
Model Size=2B, Backbon...
2026.05
60.4
SFT + GRPO
Model Size=4B, Backbon...
2026.05
57.5
SFT
Model Size=8B, Backbon...
2026.05
57.5
Base
Model Size=8B, Backbon...
2026.05
55.7
SFT
Model Size=4B, Backbon...
2026.05
51.9
SFT
Model Size=2B, Backbon...
2026.05
50
GRPO
Model Size=4B, Backbon...
2026.05
48.1
Base
Model Size=4B, Backbon...
2026.05
44.3
GRPO
Model Size=2B, Backbon...
2026.05
43.4
Base
Model Size=2B, Backbon...
2026.05
15.1
Feedback
Search any
task
Search any
task