ShowUI-π

Open-source, End-to-end, Lightweight, Vision-Language-Action Model for GUI Drag operations.

ShowUI-π 是一款开源的、端到端、轻量级的视觉-语言-动作模型，专为 GUI 拖拽交互设计。

   📑 Paper    | 🤗 Model   |    🤗 Datasets   

ShowUI-π: Flow-based Generative Models as GUI Dexterous Hands
Siyuan Hu*, Kevin Qinghong Lin*, Mike Zheng Shou
Show Lab @ National University of Singapore

🔥 Update

• [2026.2.20] ShowUI-π is accepted by CVPR 2026.
• [2025.12.31] We released ShowUI-π for GUI dragging.
• [2025.12.31] We released the DEX Benchmark for GUI drag-and-drop evaluation.

⭐ Quick Start

git clone https://github.com/showlab/showui-pi.git
cd showui-pi
pip install -e .

🚀 Training

bash scripts/train_showui_pi.sh

See scripts/train_showui_pi.sh for all flags and defaults.

🕹️ Evaluation

DEX Benchmark

The DEX benchmark is downloaded automatically on first run.

PYTHONPATH=lerobot/src \
python scripts/eval_dex.py \
    --ckpt <path/to/checkpoint> \
    --output_dir outputs/eval_dex

ScreenSpot-Pro

PYTHONPATH=lerobot/src \
python scripts/eval_screenspot_pro.py \
    --ckpt <path/to/checkpoint> \
    --annotations_root <path/to/ScreenSpot-Pro/annotations> \
    --images_root <path/to/ScreenSpot-Pro/images>

❤ Acknowledgement

We extend our gratitude to LeRobot and SmolVLA for the training framework, and SeeClick for grounding data.

🎓 BibTeX

@article{hu2025showui,
  title={ShowUI-$$\backslash$pi $: Flow-based Generative Models as GUI Dexterous Hands},
  author={Hu, Siyuan and Lin, Kevin Qinghong and Shou, Mike Zheng},
  journal={arXiv preprint arXiv:2512.24965},
  year={2025}
}

If you like our project, please give us a star ⭐ on GitHub for the latest update.