Robot Manipulation on Libero (Success Rate & Steps)
98.8Spatial Success Rateπ0.5
Evaluation results
| Method | Links | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| π0.5Acceleration Strategy=base2026.03 | 98.8 | 105.3 | 92.4 | 267.9 | 98 | 113.1 | 98.2 | 138.1 | 96.9 | 1 | — | — | |
| M2-VLAVLM Fine-tuning=No, Params (B)=0.32026.04 | 97.8 | — | 87 | — | 97.2 | — | 99 | — | 95.3 | — | — | — | |
| IVLR2026.05 | 97.8 | — | 92.4 | — | 97 | — | 95.8 | — | 95.5 | — | — | — | |
| SuPBackbone=π0.52026.03 | 97.2 | 70.4 | 94 | 215.2 | 98.6 | 93.4 | 99.4 | 83 | 97.3 | 1.35 | — | — | |
| E02025.11 | 97.2 | — | 92.2 | — | 95 | — | 99.4 | — | — | — | — | — | |
| VLA-02026.05 | 97 | — | 87.6 | — | 96.2 | — | 97.8 | — | 94.7 | — | — | — | |
| π02025.11 | 96.8 | — | 85.2 | — | 95.8 | — | 98.8 | — | — | — | — | — | |
| DemoSpeedupBackbone=π0.52026.03 | 96.4 | 88.1 | 93.2 | 221.4 | 96.8 | 88.7 | 98.8 | 114.1 | 96.3 | 1.22 | — | — | |
| π0 FAST2025.11 | 96.4 | — | 60.2 | — | 88.6 | — | 96.8 | — | — | — | — | — | |
| π0-FAST2026.05 | 96.4 | — | 60.2 | — | 88.6 | — | 96.8 | — | 85.5 | — | — | — | |
| π0.52025.11 | 95.4 | — | 89.6 | — | 97.2 | — | 98.4 | — | — | — | — | — | |
| VLA-Adapter (Frozen)VLM Fine-tuning=No, Params (B)=0.52026.04 | 94.4 | — | 82 | — | 86.8 | — | 94.2 | — | 89.4 | — | — | — | |
| FlowVLAVLM Fine-tuning=Yes, Params (B)=7.02026.04 | 93.2 | — | 72.6 | — | 91.6 | — | 95 | — | 88.1 | — | — | — | |
| SmolVLAVLM Fine-tuning=No, Params (B)=2.22026.04 | 93 | — | 77 | — | 91 | — | 94 | — | 88.8 | — | — | — | |
| VLA-AdapterAcceleration Strategy=base2026.03 | 92.2 | 99.6 | 93.6 | 255.1 | 97 | 107 | 94.2 | 136.2 | 94.2 | 1 | — | — | |
| π0.5Acceleration Strategy=ds22026.03 | 91.4 | 67.9 | 87.4 | 153.4 | 95.2 | 67.6 | 97 | 75 | 92.8 | 1.72 | — | — | |
| SuPBackbone=VLA-Adapter2026.03 | 91.2 | 77.3 | 93.4 | 204.6 | 95.6 | 74.2 | 94.4 | 91.4 | 93.7 | 1.34 | — | — | |
| ThinkActVLM Fine-tuning=Yes, Params (B)=7.02026.04 | 88.3 | — | 70.9 | — | 87.1 | — | 91.4 | — | 84.4 | — | — | — | |
| SpatialVLA2025.11 | 88.2 | — | 55.5 | — | 78.6 | — | 89.9 | — | — | — | — | — | |
| SpatialVLA2026.05 | 88.2 | — | 55.5 | — | 78.6 | — | 89.9 | — | 78.1 | — | — | — | |
| CoT-VLAVLM Fine-tuning=Yes, Params (B)=7.02026.04 | 87.5 | — | 69 | — | 87.6 | — | 91.6 | — | 81.1 | — | — | — | |
| CoT-VLA2026.05 | 87.5 | — | 69 | — | 87.6 | — | 91.6 | — | 81.1 | — | — | — | |
| MolmoActVLM Fine-tuning=Yes, Params (B)=7.02026.04 | 87 | — | 77.2 | — | 87.6 | — | 95.4 | — | 86.6 | — | — | — | |
| VLA-InfoEntropy2026.04 | 86.4 | — | 52.2 | — | 79.4 | — | 87.6 | — | 76.4 | 1.53 | 1.214 | 31.25 | |
| Spec-VLA2026.04 | 85.8 | — | 55 | — | 74.4 | — | 85 | — | 75 | 1.32 | — | — | |
| OpenVLA2026.05 | 84.9 | — | 53.7 | — | 79.2 | — | 88.4 | — | 76.5 | — | — | — | |
| OpenVLA2025.11 | 84.7 | — | 53.7 | — | 79.2 | — | 88.4 | — | — | — | — | — | |
| OpenVLAVLM Fine-tuning=Yes, Params (B)=7.02026.04 | 84.7 | — | 53.7 | — | 79.2 | — | 88.4 | — | 76.5 | — | — | — | |
| TraceVLA2025.11 | 84.6 | — | 54.1 | — | 75.1 | — | 85.2 | — | — | — | — | — | |
| OpenVLABackbone=OpenVLA, Sources=Arxiv’242025.11 | 84.4 | — | 53.2 | — | 75.6 | — | 86.6 | — | 74.95 | 1 | 100 | — | |
| OpenVLAbaseline=true2026.04 | 84.4 | — | 53.2 | — | 75.6 | — | 86.6 | — | 75 | 1 | 1.864 | 51.91 | |
| SP-VLA2026.04 | 84.4 | — | 54.2 | — | 75.4 | — | 85.6 | — | 74.9 | 1.35 | 3.1 | — | |
| Dita2025.11 | 84.2 | — | 63.8 | — | 85.4 | — | 96.3 | — | — | — | — | — | |
| VLA-CacheBackbone=OpenVLA, Sources=NeurIPS’252025.11 | 83.8 | — | 52.8 | — | 76.4 | — | 85.8 | — | 74.7 | 1.46 | 80.8 | — | |
| VLA-Cache2026.04 | 83.8 | — | 52.8 | — | 76.4 | — | 85.8 | — | 74.7 | 1.38 | 1.355 | 34.38 | |
| FastVBackbone=OpenVLA, Sources=ECCV’242025.11 | 83.4 | — | 51.6 | — | 74.2 | — | 84 | — | 73.3 | 1.11 | 41.6 | — | |
| FastV2026.04 | 83.4 | — | 51.6 | — | 74.2 | — | 84 | — | 73.3 | 1 | 1.864 | 53.28 | |
| ActDistillBackbone=OpenVLA, Sources=Ours2025.11 | 81.8 | — | 55.8 | — | 72.4 | — | 85.8 | — | 73.95 | 1.59 | 49.5 | — | |
| VLA-AdapterAcceleration Strategy=ds22026.03 | 80.2 | 57.1 | 83.4 | 147.6 | 93 | 57.1 | 88.2 | 76.3 | 86.2 | 1.77 | — | — | |
| SparseVLMBackbone=OpenVLA, Sources=ICML’252025.11 | 79.8 | — | 39.4 | — | 72.6 | — | 67 | — | 64.7 | 1.13 | 75.2 | — | |
| SparseVLM2026.04 | 79.8 | — | 39.4 | — | 72.6 | — | 67 | — | 64.7 | 1.32 | 1.407 | 83.39 | |
| Octo2025.11 | 78.9 | — | 51.1 | — | 84.6 | — | 85.7 | — | — | — | — | — | |
| Octo2026.05 | 78.9 | — | 51.1 | — | 84.6 | — | 85.7 | — | 75.1 | — | — | — | |
| MDT2025.11 | 78.5 | — | 64.8 | — | 73.5 | — | 87.5 | — | — | — | — | — | |
| Diffusion Policy2025.11 | 78.3 | — | 50.5 | — | 68.3 | — | 92.5 | — | — | — | — | — | |
| DP*2026.05 | 78.3 | — | 50.5 | — | 68.3 | — | 92.5 | — | 72.4 | — | — | — |