F Lite is a 10B parameter diffusion model created by Freepik and Fal, trained exclusively on copyright-safe and SFW content. The model was trained on Freepik's internal dataset comprising approximately 80 million copyright-safe images, making it the first publicly available model of this scale trained exclusively on legally compliant and SFW content.

F Lite Texture is a variant of F Lite that produces better textures and aesthetics in generated images, but performs poorly with short prompts and is less effective for vector graphics.

Usage

Experience F Lite Texture instantly through our interactive demo on Hugging Face of at fal.ai.

F Lite Texture works with both the diffusers library and ComfyUI. For details, see the F-Lite GitHub repository.

Technical Report

Read the technical report to learn more about the model details.

Limitations and Bias

• The models can generate malformations.
• The text capabilities of the model are limited.
• The model can be subject to biases, although we think we have a good balance given the quality and variety of the Freepik's dataset.

Recommendations

• Use long prompts to generate better results. Short prompts may result in low-quality images.
• Generate images above the megapixel. Smaller images will result in low-quality images.

Acknowledgements

This model uses T5 XXLand Flux Schnell VAE

License

The F Lite Texture weights are licensed under the permissive CreativeML Open RAIL-M license. The T5 XXL and Flux Schnell VAE are licensed under Apache 2.0.

Citation

If you find our work helpful, please cite it!

@article{ryu2025flite,
  title={F-Lite Technical Report},
  author={Ryu, Simo and Pengqi, Lu and Mart\'in Juan, Javier and de Prado Alonso, Iv\'an},
  year={2025}
}