Stable Diffusion
Stable Diffusion is a family of image models from Stability AI, rather than one free website. Downloadable weights let you run supported models on your own hardware and build a repeatable image workflow. The specific model, interface, and license determine what you can do.
What it offers
Stable Diffusion 3.5 includes Large, Large Turbo, and Medium variants. The official release provides weights and inference code and describes support for customization and fine-tuning. Local deployment suits concept art, illustration drafts, and teams that want control over processing. Editing tools depend on the selected model and application, so do not assume every checkpoint includes inpainting or outpainting.
Access and cost
For covered Core Models, the Community License permits use by individuals and organizations below US$1 million in annual revenue, subject to its terms. Larger organizations need to review enterprise licensing. This is not a blanket open-source license for every Stable Diffusion derivative. Hosted API use has separate credit billing, with one credit equal to US$0.01. Downloading weights does not provide free hosted inference.
Practical limits
Before choosing a local model, check its model card, GPU memory needs, and the licenses of any adapters or additional checkpoints. Budget for hardware, storage, and generation time. Keep model versions and generation settings with your project if later revisions need comparable results. Review images for text, anatomy, and factual errors before publication.