SmolVLA (450M)
Designed for low-cost arms (SO-100/SO-101). HF repo still updated in Sept 2026; variants lerobot/smolvla_libero, lerobot/smolvla_robotwin. NVIDIA announced it would acquire Hugging Face (see 2026-09-03 entry).
- Input
- image, text
- Output
- action
- License
- apache-2.0
- Verified
- 2026-09-29
How to call it
| Provider | Model id | Endpoint / URL | Docs |
|---|---|---|---|
| Hugging Face | lerobot/smolvla_base | huggingface.co/lerobot/smolvla_base | docs |
| GitHub (LeRobot) | — | github.com/huggingface/lerobot | — |
Notable capabilities (3)
- VLA small enough for a laptop: 450M params (SmolVLM2-500M backbone + flow-matching action expert); trains on a single GPU and runs on consumer hardware incl. MacBooks. source
- Trained on community-shared data: Pretrained on ~10M frames from 487 community LeRobot datasets (<30k episodes, an order of magnitude less than other VLAs); 78.3% success on real SO-100 tasks. source
- Asynchronous inference: Decouples action prediction from execution: ~30% faster task completion and 2x throughput. source