---
license: mit
---
<div align="center">

<p align="center">
  <img src="https://raw.githubusercontent.com/zjunlp/LabVLA/main/assets/logo/labvla-symbol.png" width="88" alt="LabVLA symbol" /><img src="https://raw.githubusercontent.com/zjunlp/LabVLA/main/assets/logo/labvla-wordmark.png" height="56" alt="LabVLA" />
</p>

<h3 align="center"> Grounding Vision–Language–Action Models in Scientific Laboratories </h3>

<p align="center">
  <a href="https://huggingface.co/papers/2606.13578">📰HF Paper</a> •
  <a href="https://zjunlp.github.io/LabVLA/">🔥Project Page</a> •
  <a href="https://github.com/zjunlp/LabVLA">💻GitHub Repo</a> •
  <a href="https://huggingface.co/zjunlp/LabVLA">🤗Model</a>
</p>
</div>

---

## Model Description

**LabVLA** is the first vision–language–action (VLA) model designed for scientific laboratory environments. It combines a **Qwen3-VL-4B-Instruct** vision–language backbone with a **DiT flow-matching action expert**, trained with the π0.5 recipe to enable real-time robot control in lab settings.

## How to Use

### Download

```bash
huggingface-cli download zjunlp/LabVLA --local-dir LabVLA
```

### Deployment

Serve the model over the OpenPI msgpack WebSocket protocol:

```bash
git clone https://github.com/zjunlp/LabVLA.git
cd LabVLA
bash deployment/deploy.sh
```

For training, data preparation, and more details, please refer to our [GitHub repository](https://github.com/zjunlp/LabVLA).