Robot Manipulation on SimplerEnv WidowX
83.3Overall Success RateCoarse-to-Control
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Coarse-to-ControlCoT Type=Action CoT2026.06 | 83.3 | 100 | 95.8 | 79.2 | 58.3 | |
| PhysBrain 1.02026.05 | 80.2 | 95.8 | 65.5 | 59.4 | 100 | |
| Xiaomi-Robotics-02026.02 | 79.2 | 95.8 | 62.5 | 75 | 83.3 | |
| LaMPMotion Guidance=Enabled2026.03 | 79.2 | 79.1 | 66.7 | 75 | 95.8 | |
| Xiaomi-Robotics-02026.05 | 79.2 | 95.8 | 62.5 | 75 | 83.3 | |
| CoWVLA2026.03 | 76 | 79.2 | 66.7 | 62.5 | 95.8 | |
| MindLWPISize=4B, Fine-tuning Protocol=Bridge-only fine-tuning2026.07 | 75.5 | — | — | — | — | |
| MindWPISize=4B, Fine-tuning Protocol=Bridge-only fine-tuning2026.07 | 74.5 | — | — | — | — | |
| FlowVLAMethod Paradigm=w/ World Model2026.03 | 74 | 70.8 | 62.5 | 62.5 | 100 | |
| FlowVLAArchitecture Category=2D Flow/Trace-Guided VLA2026.03 | 74 | 70.8 | 62.5 | 62.5 | 100 | |
| F1Architecture Category=Video-Based VLA2026.03 | 72.9 | 50 | 70.8 | 50 | 66.7 | |
| EO-12026.02 | 72.7 | 63.6 | 54.5 | 81.8 | 90.9 | |
| FutureVLA-GTStyle=GR00T-style2026.03 | 71.9 | 83.3 | 75 | 29.2 | 100 | |
| MemoryVLASize=7B, Fine-tuning Protocol=Bridge-only fine-tuning2026.07 | 71.9 | — | — | — | — | |
| MindWPI w/o PTSize=4B, Fine-tuning Protocol=Bridge-only fine-tuning2026.07 | 71.9 | — | — | — | — | |
| AsyncVLA2025.11 | 70.8 | 70.8 | 66.7 | 58.3 | 87.5 | |
| π02026.02 | 69.2 | 83.8 | 52.5 | 52.5 | 87.9 | |
| UniVLAMethod Paradigm=w/ World Model2026.03 | 68.7 | 83.3 | 62.5 | 29.2 | 100 | |
| UniVLAArchitecture Category=Latent-Action VLA2026.03 | 68.7 | 83.3 | 62.5 | 29.2 | 100 | |
| OFlow2026.04 | 67.1 | 76.8 | 62.7 | 72.4 | 56.6 | |
| MoT-HRA2026.04 | 66.1 | 78.1 | 62.5 | 40.6 | 83.3 | |
| MindPI (Full PT)Size=4B, Fine-tuning Protocol=Bridge-only fine-tuning2026.07 | 65.9 | — | — | — | — | |
| MindLPISize=4B, Fine-tuning Protocol=Bridge-only fine-tuning2026.07 | 65.6 | — | — | — | — | |
| FPC-VLASize=7B, Fine-tuning Protocol=Bridge-only fine-tuning2026.07 | 64.6 | — | — | — | — | |
| FutureVLA-OTStyle=OFT-style2026.03 | 63.6 | 75 | 54.2 | 25 | 100 | |
| villa-XMethod Paradigm=w/ Latent Actions2026.03 | 62.5 | 77.9 | 46.3 | 61.3 | 64.6 | |
| villa-XArchitecture Category=Latent-Action VLA2026.03 | 62.5 | 77.9 | 46.3 | 61.3 | 64.6 | |
| UD-VLA2025.11 | 62.5 | 58.3 | 62.5 | 54.1 | 75 | |
| UD-VLACoT Type=Visual CoT2026.06 | 62.5 | 58.3 | 62.5 | 54.1 | 75 | |
| GR00T-N1.52026.03 | 61.9 | 75.3 | 54.3 | 57 | 61.3 | |
| GR00T-N1.52026.04 | 59.8 | 71.2 | 55.6 | 61.3 | 51.2 | |
| MindPI w/o PTSize=4B, Fine-tuning Protocol=Bridge-only fine-tuning2026.07 | 59.6 | — | — | — | — | |
| FICoT Type=Visual CoT2026.06 | 59.4 | 50 | 70.8 | 50 | 66.7 | |
| Ours + Qwen3VLBackbone=Qwen3VL2026.03 | 58.3 | 66.7 | 50 | 20.8 | 95.8 | |
| LAPAMethod Paradigm=w/ Latent Actions2026.03 | 57.3 | 70.8 | 45.8 | 54.2 | 58.3 | |
| π0.52026.03 | 57.1 | 49.3 | 64.7 | 44.7 | 69.7 | |
| Isaac-GR00T-N1.6-Bridge2026.03 | 57.1 | 64.5 | 65.5 | 5.5 | 93 | |
| π0.5Architecture Category=General VLA2026.03 | 57.1 | 49.3 | 64.7 | 44.7 | 69.7 | |
| π0.52025.11 | 57.1 | 49.3 | 64.7 | 44.7 | 69.7 | |
| π0.52026.05 | 57.1 | 49.3 | 64.7 | 44.7 | 69.7 | |
| Isaac-GR00T-N1.6-BridgeArchitecture=N1.6, Training=Bridge2026.05 | 57.1 | 64.5 | 65.5 | 5.5 | 93 | |
| mimic-videoArchitecture Category=Video-Based VLA2026.03 | 56.3 | 41.7 | 54.2 | 29.2 | 100 | |
| LaMP (w/o motion)Motion Guidance=Disabled2026.03 | 56.3 | 66.7 | 45.8 | 25 | 87.5 | |
| MindPI (Frozen VLM)Size=4B, Fine-tuning Protocol=Bridge-only fine-tuning2026.07 | 54.4 | — | — | — | — | |
| VideoVLA2026.03 | 53.1 | 75 | 20.8 | 45.8 | 70.8 | |
| π02026.03 | 53.1 | 29.2 | 62.5 | 29.2 | 91.6 | |
| VideoVLA2026.05 | 53.1 | 75 | 20.8 | 45.8 | 70.8 | |
| CogACTMethod Paradigm=Standard VLA2026.03 | 51.3 | 71.7 | 50.8 | 15 | 67.5 | |
| CogACT2026.03 | 51.3 | 71.7 | 50.8 | 15 | 67.5 | |
| CogACT2026.05 | 51.3 | 71.7 | 50.8 | 15 | 67.5 | |
| CogACTCoT Type=No CoT2026.06 | 51.3 | 71.7 | 50.8 | 15 | 67.5 | |
| π0+SVAbase_policy=π02026.07 | 50.7 | 27.8 | 27.8 | 63.9 | 83.3 | |
| QwenOFT + Qwen3VLBackbone=Qwen3VL2026.03 | 50 | 58.3 | 50 | 20.8 | 70.8 | |
| GR00T N1Method Paradigm=Standard VLA2026.03 | 49.5 | 62.5 | 45.8 | 16.7 | 20.8 | |
| GR00T N1Architecture Category=General VLA2026.03 | 49.5 | 62.5 | 45.8 | 16.7 | 20.8 | |
| DD-VLASize=7B, Fine-tuning Protocol=Bridge-only fine-tuning2026.07 | 49.3 | — | — | — | — | |
| π0-FASTMethod Paradigm=Standard VLA2026.03 | 48.3 | 29.1 | 21.9 | 10.8 | 66.6 | |
| pi_0-FASTSpeed=FAST2026.04 | 48.3 | 29.1 | 21.9 | 10.8 | 66.6 | |
| TLAMethod Paradigm=w/ Latent Actions2026.03 | 48 | 52.8 | 55.6 | 2.8 | 80.6 | |
| UniVLA2026.03 | 47.9 | 52.8 | 55.6 | 2.8 | 80.6 | |
| π0+RoboMonkeybase_policy=π02026.07 | 46.3 | 29.6 | 33.3 | 40.7 | 81.5 | |
| Magma2026.02 | 44.8 | 37.5 | 29.2 | 20.8 | 91.7 | |
| Magma2026.03 | 44.8 | 37.5 | 29.2 | 20.8 | 91.7 | |
| Magma2025.11 | 44.8 | 37.5 | 29.2 | 20.8 | 91.7 | |
| ThinkAct2026.02 | 43.8 | 58.3 | 37.5 | 8.7 | 70.8 | |
| ThinkACT2026.04 | 43.8 | 58.3 | 37.5 | 8.7 | 70.8 | |
| ThinkAct2025.11 | 43.8 | 58.3 | 37.5 | 8.7 | 70.8 | |
| ThinkActCoT Type=Textual CoT2026.06 | 43.8 | 58.3 | 37.5 | 8.7 | 70.8 | |
| SpatialVLA2026.02 | 42.7 | 16.7 | 25 | 29.2 | 100 | |
| SpatialVLAMethod Paradigm=Standard VLA2026.03 | 42.7 | 16.7 | 25 | 29.2 | 100 | |
| RoboVLM2026.03 | 42.7 | 50 | 37.5 | 0 | 83.3 | |
| SpatialVLA2026.04 | 42.7 | 16.7 | 25 | 29.2 | 100 | |
| RoboVLM2026.05 | 42.7 | 50 | 37.5 | 0 | 83.3 | |
| SpatialVLASize=4B, Fine-tuning Protocol=Bridge-only fine-tuning2026.07 | 42.7 | — | — | — | — | |
| OpenVLA-OFTStrategy=OFT2026.03 | 41.8 | 34.2 | 30 | 30 | 72.5 | |
| OpenVLA-OFT2026.04 | 41.7 | 34.2 | 30 | 30 | 72.5 | |
| Villa-X2026.03 | 40.8 | 48.3 | 24.2 | 19.2 | 71.7 | |
| π0Method Paradigm=Standard VLA2026.03 | 40.1 | 29.1 | 0 | 16.7 | 62.5 | |
| π0Architecture Category=General VLA2026.03 | 40.1 | 29.1 | 0 | 16.7 | 62.5 | |
| π0CoT Type=No CoT2026.06 | 40.1 | 29.1 | 0 | 16.7 | 62.5 | |
| π0+FASTSize=3B, Fine-tuning Protocol=Bridge-only fine-tuning2026.07 | 39.5 | — | — | — | — | |
| π02026.07 | 38.5 | 23.6 | 23.6 | 37.5 | 69.4 | |
| RoboVLMs2026.02 | 37.5 | 45.8 | 20.8 | 4.2 | 79.2 | |
| RoboVLMs2026.04 | 37.5 | 45.8 | 20.8 | 4.2 | 79.2 | |
| RoboVLM2025.11 | 37.5 | 45.8 | 20.8 | 4.2 | 79.2 | |
| SpatialVLA2026.03 | 34.4 | 20.8 | 20.8 | 25 | 70.8 | |
| SpatialVLA2025.11 | 34.4 | 20.8 | 20.8 | 25 | 70.8 | |
| SpatialVLA2026.05 | 34.4 | 20.8 | 20.8 | 25 | 70.8 | |
| π0-FAST2026.02 | 32.1 | 29.1 | 21.9 | 10.8 | 66.6 | |
| π0-FAST2026.04 | 32.1 | 29.1 | 21.9 | 10.8 | 66.6 | |
| π0-FAST2025.11 | 32.1 | 29.1 | 21.9 | 10.8 | 66.6 | |
| OpenVLA-OFTSize=7B, Fine-tuning Protocol=Bridge-only fine-tuning2026.07 | 31.3 | — | — | — | — | |
| OpenVLA-OFTStyle=OFT-style2026.03 | 30.2 | 12.5 | 4.2 | 4.2 | 100 | |
| Octo-Small2026.02 | 29.5 | 47.2 | 9.7 | 4.2 | 56.9 | |
| Octo-SmallModel size=Small2025.11 | 29.5 | 47.2 | 9.7 | 4.2 | 56.9 | |
| OctoCoT Type=No CoT2026.06 | 29.5 | 47.2 | 9.7 | 4.2 | 56.9 | |
| π0Size=3B, Fine-tuning Protocol=Bridge-only fine-tuning2026.07 | 27.8 | — | — | — | — | |
| TraceVLA2026.03 | 27.7 | 12.5 | 16.6 | 16.6 | 65 | |
| TraceVLAArchitecture Category=2D Flow/Trace-Guided VLA2026.03 | 27.7 | 12.5 | 16.6 | 16.6 | 65 | |
| TraceVLA2026.05 | 27.7 | 12.5 | 16.6 | 16.6 | 65 |