gpic-baselines

GPIC Baseline Models

This repository contains the reference baseline models for GPIC: A Giant Permissive Image Corpus for Visual Generation.

GPIC is a large-scale image corpus comprising approximately 28 trillion pixels and 100M training examples, all under permissive licenses. These models provide a reference baseline for pixel-space flow matching trained on this dataset.

• Paper: GPIC: A Giant Permissive Image Corpus for Visual Generation
• Project Page: https://gpic.stanford.edu
• Code: https://github.com/keshik6/gpic
• Dataset: stanford-vision-lab/giant-permissive-image-corpus

Citation

@misc{chandrasegaran2026gpic,
      title={GPIC: A Giant Permissive Image Corpus for Visual Generation}, 
      author={Keshigeyan Chandrasegaran and Kyle Sargent and Suchir Agarwal and Michael Jang and Michael Poli and Juan Carlos Niebles and Justin Johnson and Jiajun Wu and Li Fei-Fei},
      year={2026},
      eprint={2605.30341},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2605.30341}, 
}