Robotic Manipulation on LIBERO 1.0 (test)
98.7LongQuoVLA
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| QuoVLA2026.05 | 98.7 | 99.8 | 99.9 | 100 | 99.6 | — | — | |
| X-VLA2026.05 | 97.6 | 98.2 | 98.6 | 97.8 | 98.1 | — | — | |
| ABot-M02026.05 | 96.6 | 98.8 | 99.8 | 99 | 98.6 | — | — | |
| π0.5Model Size=3.3B2026.03 | 96.5 | 99 | 97 | 99 | 97.9 | 1 | 100 | |
| SimVLA2026.05 | 96.4 | 99.6 | 99.8 | 98.6 | 98.6 | — | — | |
| Being-H0.52026.05 | 96.2 | 97 | 98.2 | 99 | 97.6 | — | — | |
| π0.5 + DepthCacheBackbone=π0.5 (3.3B), Compression=DepthCache2026.03 | 96.1 | 98.9 | 97.1 | 98.3 | 97.6 | 1.28 | 68.2 | |
| VOTE2026.05 | 95.6 | 98.8 | 99.8 | 97.6 | 98 | — | — | |
| FASTerArchitecture=Autoregressive, BAR=Yes2025.12 | 95.4 | 98 | 99.4 | 98.6 | 97.9 | — | — | |
| VLA-Adapter2026.05 | 95 | 97.8 | 99.2 | 97.2 | 97.3 | — | — | |
| OpenVLA-OFTArchitecture=Non-autoregressive2025.12 | 94.5 | 97.6 | 98.4 | 97.9 | 97.1 | — | — | |
| OpenVLA-OFT2026.05 | 94.5 | 97.6 | 98.4 | 97.9 | 97.1 | — | — | |
| FLOWER2026.05 | 93.5 | 97.1 | 96.7 | 95.6 | 95.7 | — | — | |
| MemoryVLA2026.05 | 93.4 | 98.4 | 98.4 | 96.4 | 96.7 | — | — | |
| π05Architecture=Non-autoregressive2025.12 | 92.4 | 98.8 | 98.2 | 98 | 96.8 | — | — | |
| pi0.52026.05 | 92.4 | 98.8 | 98.2 | 98 | 96.9 | — | — | |
| UniVLAArchitecture=Non-autoregressive2025.12 | 92 | 96.5 | 96.8 | 95.6 | 95.2 | — | — | |
| UniVLA2026.05 | 92 | 96.5 | 96.8 | 95.6 | 95.2 | — | — | |
| GR00TModel Size=2.2B2026.03 | 90.2 | 91.7 | 97.6 | 93 | 93.1 | 1 | 100 | |
| GR00T + DepthCacheBackbone=GR00T (2.2B), Compression=DepthCache2026.03 | 89.6 | 93.3 | 96.4 | 92.5 | 92.9 | 1.07 | 87.5 | |
| DreamVLA2025.07 | 89.5 | 97.5 | 94 | 89.5 | 92.6 | — | — | |
| FASTerArchitecture=Autoregressive, BAR=No2025.12 | 88.6 | 99.4 | 98.8 | 94.8 | 95.4 | — | — | |
| COT-VLA2025.07 | 87.6 | 81.1 | 87.5 | 91.6 | 69 | — | — | |
| π0 FAST-DArchitecture=Autoregressive, Action Chunks=Delta2025.12 | 86.8 | 96.6 | 97.2 | 96 | 94.2 | — | — | |
| π0Architecture=Non-autoregressive2025.12 | 85.2 | 96.8 | 98.8 | 95.8 | 94.2 | — | — | |
| pi02026.05 | 85.2 | 96.8 | 98.8 | 95.8 | 94.2 | — | — | |
| Ditainput=with wrist camera2025.03 | 83.6 | 97.4 | 94.8 | 93.2 | 92.3 | — | — | |
| MiniVLAArchitecture=Autoregressive2025.12 | 77 | — | — | — | — | — | — | |
| CoT-VLAModel Scale=7B2025.03 | 69 | 87.5 | 91.6 | 87.6 | 81.13 | — | — | |
| CoT-VLA2026.05 | 69 | 87.5 | 91.6 | 87.6 | 83.9 | — | — | |
| π0.5 + FastVBackbone=π0.5 (3.3B), Compression=FastV2026.03 | 68.5 | 73.4 | 89.4 | 79.1 | 77.6 | 1.3 | 68 | |
| Diffusion Policy2026.05 | 64.8 | 78.5 | 87.5 | 73.5 | 76.1 | — | — | |
| Dita2025.03 | 63.8 | 84.2 | 96.3 | 85.4 | 82.4 | — | — | |
| π0.5 + ToSABackbone=π0.5 (3.3B), Compression=ToSA2026.03 | 62.4 | 79.3 | 82 | 71.6 | 73.8 | 0.94 | 57.8 | |
| π0 FAST-RArchitecture=Autoregressive, Action Chunks=Relative2025.12 | 60.2 | 96.4 | 96.8 | 88.6 | 85.5 | — | — | |
| VQ-VLAArchitecture=Autoregressive2025.12 | 60 | — | — | 75.2 | — | — | — | |
| SpatialVLA2025.07 | 55.5 | 88.2 | 89.9 | 78.6 | 78.1 | — | — | |
| SpatialVLAArchitecture=Non-autoregressive2025.12 | 55.5 | 88.2 | 89.9 | 78.6 | 78.1 | — | — | |
| SpatialVLA2026.05 | 55.5 | 88.2 | 89.9 | 78.6 | 78.1 | — | — | |
| OpenVLA2025.07 | 53.7 | 84.7 | 88.4 | 79.2 | 76.5 | — | — | |
| OpenVLA2025.03 | 53.7 | 84.9 | 88.4 | 79.2 | 76.5 | — | — | |
| OpenVLAmode=fine-tuned2025.03 | 53.7 | 84.7 | 88.4 | 79.2 | 76.5 | — | — | |
| OpenVLAArchitecture=Autoregressive2025.12 | 53.7 | 84.7 | 88.4 | 79.2 | 76.5 | — | — | |
| OpenVLAModel Size=7B2026.03 | 53.7 | 83.3 | 91.2 | 78.6 | 76.7 | 1 | 100 | |
| OpenVLA2026.05 | 53.7 | 84.7 | 88.4 | 79.2 | 76.5 | — | — | |
| OpenVLA + SP-VLABackbone=OpenVLA (7B), Compression=SP-VLA2026.03 | 51.6 | 80 | 82.4 | 73.6 | 71.9 | 1.5 | — | |
| OpenVLA + DepthCacheBackbone=OpenVLA (7B), Compression=DepthCache2026.03 | 51.2 | 83.1 | 90.7 | 77.8 | 75.7 | 1.21 | 78.9 | |
| Octo2025.07 | 51.1 | 78.9 | 85.7 | 84.6 | 75.1 | — | — | |
| Octo2025.03 | 51.1 | 78.9 | 85.7 | 84.6 | 75.1 | — | — | |
| Octomode=fine-tuned2025.03 | 51.1 | 78.9 | 85.7 | 84.6 | 75.1 | — | — | |
| Octo-BaseArchitecture=Non-autoregressive2025.12 | 51.1 | 78.9 | 85.7 | 84.6 | 75.1 | — | — | |
| Diffusion Policy2025.07 | 50.5 | 78.3 | 92.5 | 68.3 | 72.4 | — | — | |
| DP*training=from scratch2025.03 | 50.5 | 78.3 | 92.5 | 68.3 | 72.4 | — | — | |
| Diffusion Policy2025.03 | 50.5 | 78.3 | 92.5 | 68.3 | 72.4 | — | — | |
| Diffusion PolicyArchitecture=Non-autoregressive2025.12 | 50.5 | 78.3 | 92.5 | 68.3 | 72.4 | — | — | |
| OpenVLA + FastVBackbone=OpenVLA (7B), Compression=FastV2026.03 | 42.5 | 75.8 | 70.3 | 67.3 | 64 | 1.39 | 75 | |
| Palligemma + Naive TokenizerArchitecture=Autoregressive2025.12 | 31.2 | 55.8 | 64.8 | 64.4 | 54.1 | — | — |