Robot Manipulation on SimplerEnv Google Robot Visual Matching
98.7Pick Coke CanXiaomi-Robotics-0
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Xiaomi-Robotics-02026.02 | 98.7 | 88.8 | 79.6 | 75 | 85.5 | — | — | |
| Xiaomi-Robotics-02026.02 | 98.7 | 88.8 | 79.6 | 75 | 85.5 | — | — | |
| Villa-x2025.10 | 98.7 | 75 | 59.3 | — | — | 5.6 | 59.6 | |
| UniJEPA2025.10 | 98.7 | 81.5 | 63.2 | — | — | 70 | 78.4 | |
| EO-12026.02 | 98 | 83.8 | 71.3 | 52.8 | 76.5 | — | — | |
| EO-12026.02 | 98 | 83.8 | 71.3 | 52.8 | 76.5 | — | — | |
| π02026.02 | 97.9 | 78.7 | 62.3 | 46.6 | 71.4 | — | — | |
| π02026.02 | 97.9 | 78.7 | 62.3 | 46.6 | 71.4 | — | — | |
| StarVLA-OFTbackbone=Qwen3-VL-4B2026.04 | 95.3 | 75 | 68.8 | — | — | 66.1 | 76 | |
| π0reproduced=true2025.10 | 93.3 | 78.1 | 23.6 | — | — | 12.5 | 51.9 | |
| DA-PTQSources=—2026.04 | 92.4 | 87.9 | 58.3 | — | — | 35.2 | 68.5 | |
| ThinkAct2026.02 | 92 | 72.4 | 50 | — | — | — | — | |
| ThinkAct2026.02 | 92 | 72.4 | 50 | — | — | — | — | |
| VLA-CacheSources=NeurIPS’252026.04 | 92 | 83.3 | 70.5 | — | — | 51.6 | 74.4 | |
| CogACT2026.04 | 91.3 | 85 | 71.8 | — | — | 50.9 | 74.8 | |
| CogACT (FP)Sources=Arxiv’242026.04 | 91.3 | 85 | 71.8 | — | — | 50.9 | 74.8 | |
| CogACT2025.10 | 91.3 | 85 | 71.8 | — | — | 50.9 | 74.8 | |
| QuantVLASources=CVPR’262026.04 | 87.6 | 81.7 | 55.1 | — | — | 38 | 65.6 | |
| SpatialVLA2026.02 | 86 | 77.9 | 57.4 | 0 | 55.3 | — | — | |
| SpatialVLA2026.02 | 86 | 77.9 | 57.4 | 0 | 55.3 | — | — | |
| SpatialVLA2026.04 | 86 | 77.9 | 57.4 | — | — | — | 75.1 | |
| RT-12026.02 | 85.7 | 44.2 | 73 | 6.5 | 52.4 | — | — | |
| RT-12026.02 | 85.7 | 44.2 | 73 | 6.5 | 52.4 | — | — | |
| RT-12026.04 | 85.7 | 44.2 | 73 | — | — | 6.5 | 52.4 | |
| SOMAPre-training Dataset=OXE, Fine-tuning Dataset=Fractal2026.05 | 85 | 73 | 31.5 | — | — | — | 63.2 | |
| π0.5 + SG-LPState Grounded Latent Prompt=true2026.06 | 85 | 66.2 | 55.6 | — | — | — | 68.9 | |
| TTT-VLAState Grounded Latent Prompt=true, Test-Time Training=true2026.06 | 85 | 71.7 | 60.6 | — | — | — | 72.4 | |
| π0.5Description=baseline fine-tuned from original checkpoint2026.06 | 84 | 59.2 | 59.3 | — | — | — | 67.5 | |
| Magma2026.04 | 83.7 | 65.4 | 56 | — | — | 6.4 | 52.9 | |
| RT-2-X2026.02 | 78.7 | 77.9 | 25 | 7.4 | 47.3 | — | — | |
| RT-2-X2026.02 | 78.7 | 77.9 | 25 | 7.4 | 47.3 | — | — | |
| RT-2-X2026.04 | 78.7 | 77.9 | 25 | — | — | 3.7 | 46.3 | |
| MolmoAct2026.02 | 77.7 | 77.1 | 60 | — | — | — | — | |
| MolmoAct2026.02 | 77.7 | 77.1 | 60 | — | — | — | — | |
| RoboVLMs2026.02 | 77.3 | 61.7 | 43.5 | 24.1 | 51.7 | — | — | |
| RoboVLMs2026.02 | 77.3 | 61.7 | 43.5 | 24.1 | 51.7 | — | — | |
| RoboVLMs2025.10 | 77.3 | 61.7 | 43.5 | — | — | 24.1 | 51.7 | |
| RoboVLMPre-training Dataset=OXE, Fine-tuning Dataset=Fractal2026.05 | 77.3 | 61.7 | 43.5 | — | — | — | 60.6 | |
| RoboVLMProtocol=fine-tuned2026.06 | 77.3 | 61.7 | 43.5 | — | — | — | 63.4 | |
| π0-FAST2026.02 | 75.3 | 67.5 | 42.9 | 0 | 46.4 | — | — | |
| π0-FAST2026.02 | 75.3 | 67.5 | 42.9 | 0 | 46.4 | — | — | |
| π0-FAST2026.04 | 75.3 | 67.5 | 42.9 | — | — | — | 61.9 | |
| π0-Fast2025.10 | 75.3 | 67.5 | 42.9 | — | — | 62 | 61.9 | |
| π0+FASTPre-training Dataset=OXE, Fine-tuning Dataset=Fractal2026.05 | 75.3 | 67.5 | 42.9 | — | — | — | 61.9 | |
| π0-FastProtocol=fine-tuned2026.06 | 75.3 | 67.5 | 42.9 | — | — | — | 61.9 | |
| Magma2026.02 | 75 | 53 | 58.9 | 8.3 | 48.8 | — | — | |
| Magma2026.02 | 75 | 53 | 58.9 | 8.3 | 48.8 | — | — | |
| MotoPre-training Dataset=OXE, Fine-tuning Dataset=Fractal2026.05 | 74 | 60.4 | 43.1 | — | — | — | 59.2 | |
| π02026.04 | 72.7 | 65.3 | 38.3 | — | — | — | 58.8 | |
| π0Pre-training Dataset=OXE, Fine-tuning Dataset=Fractal2026.05 | 72.7 | 65.3 | 38.3 | — | — | — | 58.8 | |
| RoboVLMProtocol=zero-shot2026.06 | 72.7 | 66.3 | 26.8 | — | — | — | 56.3 | |
| π0Protocol=fine-tuned2026.06 | 72.7 | 65.3 | 38.3 | — | — | — | 58.7 | |
| OpenVLA-OFT2025.10 | 72.3 | 69.6 | 47.2 | — | — | 62.9 | 63 | |
| OpenVLA-OFTPre-training Dataset=OXE, Fine-tuning Dataset=Fractal2026.05 | 72.3 | 69.6 | 47.2 | — | — | — | 63 | |
| GR00T-N1.52025.10 | 69.3 | 68.7 | 35.8 | — | — | 60 | 57.9 | |
| GR00TProtocol=fine-tuned2026.06 | 69.3 | 68.7 | 35.8 | — | — | — | 52.4 | |
| RT-1-X2026.02 | 56.7 | 31.7 | 59.7 | 40.7 | 47.2 | — | — | |
| RT-1-X2026.02 | 56.7 | 31.7 | 59.7 | 40.7 | 47.2 | — | — | |
| RT-1-X2026.04 | 56.7 | 31.7 | 59.7 | — | — | 21.3 | 42.4 | |
| RT-1-X2025.10 | 56.7 | 31.7 | 59.7 | — | — | 21.3 | 42.4 | |
| RT-1-X2026.06 | 56.7 | 31.7 | 59.7 | — | — | — | 53.4 | |
| HPT2026.06 | 56 | 60 | 24 | — | — | — | 46 | |
| DeFI2026.03 | 54.2 | 60.7 | 38.6 | — | — | — | 51.2 | |
| GR00T N1.5status=reimplementation2026.04 | 51.7 | 54 | 27.8 | — | — | 7.4 | 35.2 | |
| GR00T-N1.5Pre-training Dataset=OXE, Fine-tuning Dataset=Fractal2026.05 | 47 | 70 | 18.1 | — | — | — | 45 | |
| TraceVLA2026.03 | 28 | 53.7 | 57 | — | — | — | 42 | |
| TraceVLAPre-training Dataset=OXE, Fine-tuning Dataset=Fractal2026.05 | 28 | 53.7 | 57 | — | — | — | 42 | |
| TraceVLA2026.06 | 28 | 53.7 | 57 | — | — | — | 42 | |
| OpenVLA2026.04 | 18 | 56.3 | 63 | — | — | 0 | 34.3 | |
| Octo-Base2026.02 | 17 | 4.2 | 22.7 | 0 | 11 | — | — | |
| Octo-Base2026.02 | 17 | 4.2 | 22.7 | 0 | 11 | — | — | |
| Octo-Base2026.03 | 17 | 4.2 | 22.7 | — | — | — | 16.8 | |
| Octo-Base2025.10 | 17 | 4.2 | 22.7 | — | — | 0 | 11 | |
| Octo-Base2026.06 | 17 | 4.2 | 22.7 | — | — | — | 16.8 | |
| OpenVLA2026.02 | 16.3 | 46.2 | 35.6 | 0 | 24.5 | — | — | |
| OpenVLA2026.02 | 16.3 | 46.2 | 35.6 | 0 | 24.5 | — | — | |
| OpenVLA2026.03 | 16.3 | 46.2 | 35.6 | — | — | — | 27.7 | |
| OpenVLA2026.06 | 16.3 | 46.2 | 35.6 | — | — | — | 27.7 | |
| Emma-X2026.06 | 2.3 | 3.3 | 18.3 | — | — | — | 8 |