LeRobot's port of the Physical Intelligence pi0.5 base model, which is co-trained on web and multi-robot data to generalize better to unfamiliar homes and objects.
lerobot/smolvla_base
Indexed from Hugging FaceLinked to its home on Hugging Face. FrontierOS indexes it; nothing is hosted or mirrored here.ModelVision-language-action450M paramsEst. Edge GPUEstimated minimum compute for inference: Jetson Orin class, 8–16 GB shared memory. Not a benchmark.Apache 2.0Single arm
About
A compact 450M-parameter VLA from Hugging Face, pretrained on community-collected LeRobot datasets and sized for fine-tuning and inference on consumer GPUs.
Why it matters for physical AI
It is one of the few VLAs light enough to fine-tune and run for low-cost arms like the SO-100/SO-101 without datacenter hardware, and it uses a permissive license.
Summary written by FrontierOS in our own words. The upstream card and license govern use. Metadata checked 27 Sep 2026. Read the full card on Hugging Face
Files, weights and version history live on Hugging Face. FrontierOS indexes this repo; it does not host or mirror it.
hf download lerobot/smolvla_baseUses the Hugging Face CLI (pip install -U huggingface_hub). This downloads the whole repo; add --include "<pattern>" to fetch only the files you need.
Fine-tuning SmolVLA on an SO-101: what's your recipe?
The SO-101 pick-and-place set is a good template: about fifty episodes with a side and a top-down camera. Most people seem to start from the SmolVLA base checkpoint.
More vision-language-action
Browse allA 3B cross-embodiment robot foundation model from NVIDIA that combines a vision-language backbone with a diffusion-transformer action head and can be post-trained on a new robot with a modest amount of data.
A 7B vision-language-action model on a Llama 2 backbone that maps one camera image and a text instruction to discretized 7-DoF end-effector actions, trained on Open X-Embodiment episodes.