Robot manipulation on Simpler-WidowX
73.7Success RateQwen-VLA-Instruct
Evaluation results
| Method | Links | |
|---|---|---|
| Qwen-VLA-InstructType=Generalist2026.05 | 73.7 | |
| StarVLA-OFTType=Specialist2026.05 | 64.6 | |
| Qwen-VLA-BaseType=Generalist2026.05 | 64.3 | |
| GR00T N1.6Type=Specialist2026.05 | 63.2 | |
| DRIFTBase Model=Qwen3-VL-2B [2], Action Decoder=DRIFT-MLP2026.06 | 61.5 | |
| VLM4VLABase Model=InternVL3.5-4B [55], Action Decoder=MLP2026.06 | 57.3 | |
| π0.5Base Model=PaliGemma-3B [4], Action Decoder=Flow2026.06 | 57.1 | |
| InternVLABase Model=Qwen2.5-VL-3B [3], Action Decoder=Diffusion2026.06 | 54.2 | |
| VLM4VLABase Model=Qwen3-VL-2B [2], Action Decoder=MLP2026.06 | 49 | |
| π0.5Type=Specialist2026.05 | 46.9 | |
| π0Base Model=PaliGemma-3B [4], Action Decoder=Flow2026.06 | 27.1 | |
| DRIFTBase Model=OpenVLA-7B [20], Action Decoder=DRIFT-Tokenizer2026.06 | 13.5 | |
| OpenVLA*Base Model=Prismatic-7B [17], Action Decoder=Tokenizer2026.06 | 7.7 |