Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Embodied Manipulation on Guava Simulation (ID)
Loading...
100
Success Rate: Place Can in Box
GPT-5.4
16.8
38.4
60
81.6
Jun 16, 2026
Success Rate: Place Can in Box
Overall Success Rate
Success Rate: Arrange by Size
Success Rate: Remove Cube from Tray
Success Rate: Push Basket
Success Rate: Close Drawer
Updated 1mo ago
Evaluation Results
Method
Method
Links
Success Rate: Place Can in Box
Overall Success Rate
Success Rate: Arrange by Size
Success Rate: Remove Cube from Tray
Success Rate: Push Basket
Success Rate: Close Drawer
GPT-5.4
Episodes=15
2026.06
100
-
40
86.7
26.7
53.3
Guava-Agent-4B
Episodes=15
2026.06
100
-
66.7
100
60
86.7
CaP-Agent0
Episodes=15
2026.06
86.7
-
46.7
73.3
20
67.7
Qwen3.5-4B
Episodes=15
2026.06
20
-
0
40
0
20
Feedback
Search any
task
Search any
task