Fast Stable Diffusion is a hugely popular collection of optimized notebooks that let you run Stable Diffusion — and train it with DreamBooth — in the cloud in minutes. Instead of spending an hour installing dependencies, the notebooks boot AUTOMATIC1111 or ComfyUI pre-configured and dramatically faster. The project has earned about 7,900 stars on GitHub and was still receiving updates in late 2025.
The repository removes the friction from browser-based image generation. Its notebooks pre-install and cache the heavy dependencies, load models quickly from Google Drive or Hugging Face, and expose the full interfaces of AUTOMATIC1111’s WebUI and ComfyUI, so everything you know about prompts, samplers, ControlNet-style workflows and extensions transfers directly. The DreamBooth notebooks are the other half of its popularity: with a handful of photos you can fine-tune a model on a face, a pet or a style, using sensible defaults that work on a single rented GPU. Support spans the model generations the community actually uses — SD 1.5, SDXL and FLUX — and the project became a default recommendation for anyone who wants serious Stable Diffusion without owning a high-end graphics card.
The notebooks are free and open source under the MIT license. Running them uses your own Google Colab quota or a paid Colab plan if you choose one.
Fast Stable Diffusion is for artists, hobbyists and researchers who want full-power Stable Diffusion and DreamBooth training in the cloud without local hardware or installation pain.