Visual Matching on SimplerEnv WidowX
65.7Average Success RateNoTVLA
Evaluation results
| Method | Links | |
|---|---|---|
| NoTVLABackbone / Family=Qwen3-VL-4B + narrative sparse action2025.10 | 65.7 | |
| Qwen-OFTBackbone / Family=Qwen3-VL-4B + chunked continuous2025.10 | 64.6 | |
| GR00T-N1.5Backbone / Family=dual-system VLA2025.10 | 61.9 | |
| Qwen-PIBackbone / Family=Qwen3-VL-4B + flow expert2025.10 | 60.9 | |
| CogACTBackbone / Family=hybrid policy2025.10 | 51.3 | |
| π0-FASTBackbone / Family=tokenized π variant2025.10 | 48.3 | |
| SpatialVLABackbone / Family=spatial token VLA2025.10 | 42.7 | |
| MagmaBackbone / Family=multimodal foundation model2025.10 | 35.8 | |
| π0Backbone / Family=VLA + flow expert2025.10 | 27.1 | |
| OpenVLABackbone / Family=token VLA2025.10 | 4.2 |