huggingface/diffusers

What is it?

Key features

• 01Pipelines load pretrained diffusion checkpoints from the Hugging Face Hub in a few lines of code.
• 02Interchangeable noise schedulers trade off generation speed against output quality.
• 03Pretrained models can be combined with schedulers to build custom diffusion systems.
• 04Covers text-to-image, image-to-image, inpainting, super-resolution and text-to-video tasks.
• 05Includes guides for inference optimization and training diffusion models.

Use cases

• Generating images from text prompts with Stable Diffusion or Flux checkpoints.
• Building a custom diffusion pipeline from individual models and schedulers.
• Training or fine-tuning a diffusion model for a specific image or video task.
• Adding image editing features such as inpainting or image-to-image to an application.

Quick start

1. 1

   Create a virtual environment and install Diffusers with PyTorch from PyPI

   pip install --upgrade diffusers[torch]
2. 2

   Load a pretrained pipeline and generate an image from text in Python

   from diffusers import DiffusionPipeline

Things to know

• !Generating images with the example models needs a GPU for practical speed.
• !Large pretrained checkpoints must be downloaded from the Hugging Face Hub, which takes significant disk space.

Topics

FAQ

What is diffusers?

A PyTorch library of pretrained diffusion models, noise schedulers and building blocks for image, video and audio generation. Diffusers is a Hugging Face library for running and training diffusion models. Its pipelines let you load a pretrained checkpoint from the Hugging Face Hub with a few lines of code and generate outputs from text, images or video inputs. Interchangeable noise schedulers control the trade-off between generation speed and output quality.

Is diffusers free to use? What license does it have?

diffusers is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.

Is diffusers still maintained?

Actively maintained. The last commit to diffusers was on Oct 20, 2026.

What can I use diffusers for?

Generating images from text prompts with Stable Diffusion or Flux checkpoints. Building a custom diffusion pipeline from individual models and schedulers. Training or fine-tuning a diffusion model for a specific image or video task. Adding image editing features such as inpainting or image-to-image to an application.

How do I install diffusers?

Create a virtual environment and install Diffusers with PyTorch from PyPI: pip install --upgrade diffusers[torch] Load a pretrained pipeline and generate an image from text in Python: from diffusers import DiffusionPipeline

Where can I get diffusers?

The source code and installation instructions are on GitHub: https://github.com/huggingface/diffusers