Robot Manipulation on Simulation held-out environments
92.8Pick Success Rate (MSProc)MolmoBot-Img
Evaluation Results
| Method | Links | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| MolmoBot-ImgZero-shot=true, Visual Backbone=Img2026.03 | 92.8 | 67 | 63 | 62.5 | 67 | 47.5 | 65 | 66.4 | — | — | — | — | — | — | |
| MolmoBot (F=2)Zero-shot=true, Number of frames (F)=22026.03 | 92 | 63 | 66 | 61 | 60 | 55 | 68 | 66.4 | — | — | — | — | — | — | |
| MolmoBot (F=3)Zero-shot=true, Number of frames (F)=32026.03 | 91.9 | 61 | 63.5 | 60 | 62.5 | 46.5 | 65.5 | 64.4 | — | — | — | — | — | — | |
| MolmoBot-Pi0Zero-shot=true, Visual Backbone=Pi02026.03 | 65.5 | 30.5 | 32 | 36 | 41.5 | 41.8 | 42.5 | 41.4 | — | — | — | — | — | — | |
| π0.5-FinetuneFine-tuning=true2026.03 | 47 | 20.5 | 24 | 28.5 | 46 | 36.9 | 53 | 36.6 | — | — | — | — | — | — | |
| π0.5Zero-shot=true2026.03 | 36.4 | 9.3 | 14 | 12.8 | 20 | 35.5 | 16.2 | 20.6 | — | — | — | — | — | — | |
| MolmoAct2-DROID2026.05 | 35.6 | 18.9 | 20.5 | 15.4 | — | — | — | 20.6 | 15.8 | 7.8 | 20.9 | 14.4 | 17.2 | 8.8 | |
| LAP-VLA2026.05 | 19.4 | 2.4 | 3.1 | 2.7 | — | — | — | 4.8 | 3.81 | 1.59 | 6.48 | 3.41 | 3.1 | 1.5 | |
| π0.5-DROID2026.05 | 18.1 | 6.4 | 7 | 8 | — | — | — | 10 | 11.7 | 7.6 | 8.2 | 6.2 | 10.4 | 6.7 | |
| LAP-VLAZero-shot=true2026.03 | 12.6 | 0.5 | 0.5 | 0 | 2.2 | 10.1 | 0 | 3.7 | — | — | — | — | — | — | |
| StereoVLA2026.05 | 6.6 | 4.3 | 1.1 | — | — | — | — | — | — | — | — | — | — | — | |
| StereoVLAZero-shot=true2026.03 | 6.3 | 4.3 | 1.1 | — | 0 | 0 | 0 | — | — | — | — | — | — | — | |
| X-VLAZero-shot=true2026.03 | 6 | 5.8 | 0.9 | 0.2 | 0 | 0 | 0 | 1.8 | — | — | — | — | — | — | |
| X-VLA2026.05 | 3.3 | 0.5 | 0.7 | 0.8 | — | — | — | 1.2 | 0.1 | 0.1 | 1.9 | 1 | 0.9 | 0.5 |