Long-horizon robotic manipulation on CALVIN ABC-D
0.27Average Trajectory Length3DDP
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| 3DDPTask=ABC→D2025.10 | 0.27 | 28.3 | 2.3 | 0 | 0 | 0 | |
| MCILTask=ABC→D2025.10 | 0.31 | 30.4 | 1.3 | 0.2 | 0 | 0 | |
| Diffusion PolicyZero-shot evaluation=true, External robotic data usage=false2025.09 | 0.56 | 40 | 12 | 3 | 1 | 0 | |
| RT-1Task=ABC→D2025.10 | 0.9 | 53.3 | 22.2 | 9.4 | 3.8 | 1.3 | |
| GR00T-N1.52025.10 | 0.98 | 55.8 | 25.9 | 10.7 | 4.3 | 1.3 | |
| MDT2025.10 | 1.55 | 63.1 | 42.9 | 24.7 | 15.1 | 9.1 | |
| Robo-FlamingoTask=ABC→D2025.10 | 2.47 | 82.4 | 61.9 | 46.6 | 33.1 | 23.5 | |
| Robo-FlamingoZero-shot evaluation=true, External robotic data usage=false2025.09 | 2.47 | 82 | 62 | 47 | 33 | 24 | |
| UP-VLA2025.10 | 2.74 | — | — | — | — | — | |
| Enhanced DPZero-shot evaluation=true, External robotic data usage=false2025.09 | 2.78 | 82 | 67 | 53 | 41 | 35 | |
| GR-1Task=ABC→D2025.10 | 3.06 | 85.4 | 71.2 | 59.6 | 49.7 | 40.1 | |
| GR-1Additional Data=Ego4D2025.09 | 3.06 | 85 | 71 | 60 | 50 | 40 | |
| OpenVLA2026.06 | 3.27 | 91.3 | 77.8 | 62 | 52.1 | 43.5 | |
| 3D Diffuser ActorTask=ABC→D2025.10 | 3.35 | 93.8 | 80.3 | 66.2 | 53.3 | 41.2 | |
| MoDEZero-shot evaluation=true, External robotic data usage=false2025.09 | 3.39 | 92 | 79 | 67 | 56 | 45 | |
| VidmanAdditional Data=OpenX subsets2025.09 | 3.42 | 92 | 76 | 68 | 59 | 47 | |
| SeerZero-shot evaluation=true, External robotic data usage=false2025.09 | 3.64 | 93 | 82 | 72 | 63 | 53 | |
| Seer2025.10 | 3.64 | 93 | 82.4 | 72.3 | 62.6 | 53.3 | |
| RoboUniviewZero-shot evaluation=true, External robotic data usage=false2025.09 | 3.65 | 94 | 84 | 73 | 62 | 51 | |
| FLAREBase model=GR00T-N1.5, reproduced=true2025.10 | 3.75 | 96 | 86.1 | 74.8 | 63.8 | 54.4 | |
| UniVLA2026.06 | 3.8 | 95.5 | 85.8 | 75.4 | 66.9 | 56.5 | |
| LTMAdditional Data=OpenX subsets2025.09 | 3.81 | 97 | 82 | 73 | 67 | 61 | |
| π0.52026.03 | 3.83 | 93.2 | 84.6 | 76.7 | 68.8 | 61.4 | |
| Seer-LargeZero-shot evaluation=true, External robotic data usage=false2025.09 | 3.83 | 93 | 85 | 76 | 69 | 60 | |
| TempoFitBackbone=π0.52026.03 | 3.87 | 93 | 84.8 | 77.3 | 69.4 | 62 | |
| π0Task=ABC→D2026.03 | 3.87 | 94.3 | 87 | 77.9 | 68.5 | 59.4 | |
| DUSTBase model=GR00T-N1.52025.10 | 3.91 | 93.8 | 86.5 | 78.2 | 70.5 | 62.3 | |
| VPPZero-shot evaluation=true, External robotic data usage=false2025.09 | 3.93 | 96 | 88 | 78 | 71 | 60 | |
| SeerAdditional Data=DROID2025.09 | 3.98 | 94 | 87 | 80 | 72 | 64 | |
| DAWNZero-shot evaluation=true, External robotic data usage=false2025.09 | 4 | 98 | 91 | 79 | 71 | 61 | |
| MoDEAdditional Data=Multiple sources2025.09 | 4.01 | 96 | 89 | 81 | 72 | 65 | |
| π0.5Task=ABC→D2026.03 | 4.02 | 91.9 | 84.6 | 79.4 | 75.5 | 71 | |
| UP-VLATask=ABC→D2025.10 | 4.08 | 92.8 | 86.5 | 81.5 | 76.9 | 69.9 | |
| AtomicVLATask=ABC→D2026.03 | 4.09 | 95 | 87.8 | 81.9 | 75 | 69.1 | |
| DAWNAdditional Data=DROID2025.09 | 4.1 | 98 | 92 | 81 | 75 | 64 | |
| RoboVLMTask=ABC→D2025.10 | 4.25 | 98 | 93.6 | 85.4 | 77.8 | 70.4 | |
| AtomicVLA*Task=ABC→D2026.03 | 4.27 | 94.1 | 88.7 | 85.2 | 81.7 | 77.6 | |
| Seer-LargeTask=ABC→D2025.10 | 4.28 | 96.3 | 91.6 | 86.1 | 80.3 | 74 | |
| Seer-LargeAdditional Data=DROID2025.09 | 4.28 | 96 | 92 | 86 | 80 | 74 | |
| OpenVLA-OFT2026.06 | 4.28 | 96.9 | 92 | 85.7 | 80.4 | 72.9 | |
| Seer2026.06 | 4.28 | 96.3 | 91.6 | 86.1 | 80.3 | 74 | |
| VPPAdditional Data=Multiple sources2025.09 | 4.33 | 97 | 91 | 87 | 82 | 77 | |
| FALCONTask=ABC→D2025.10 | 4.4 | 98.4 | 94.5 | 88.6 | 82.5 | 75.5 | |
| UnifiedVLA2026.06 | 4.41 | 98.9 | 94.8 | 89 | 82.8 | 75.1 | |
| VLA-Adapter2026.06 | 4.42 | 99.1 | 94.6 | 88.8 | 82.8 | 76.5 | |
| DreamVLAAdditional Data=DROID2025.09 | 4.44 | 98 | 95 | 90 | 83 | 78 | |
| FLOWER2026.06 | 4.53 | 99.4 | 95.8 | 90.7 | 84.9 | 77.8 | |
| AVA-VLA2026.06 | 4.77 | 99.7 | 96.5 | 94.5 | 91.1 | 84 |