Stable Diffusion
Open-weight image models you can run locally — no per-image fees
About
Stable Diffusion is Stability AI's family of open-weight image models (SD 1.5, SDXL, SD 3.5, and successors) that generate images from text prompts. Because the weights are public, you can run the models on your own GPU with tools like ComfyUI, fine-tune them, and build custom workflows with no per-image cost. It is best for developers and creators who want local, private, and customizable image generation.
There is no hosted version — you run it locally. The standard open-source setup is ComfyUI, which natively supports Stable Diffusion models; install it with comfy-cli, then download a Stable Diffusion checkpoint and place it in ComfyUI's models/checkpoints folder:
pip install comfy-cli
comfy installCopy individual commands
pip install comfy-clicomfy install
ComfyUI also offers a desktop app for Windows and macOS from comfy.org, and Windows portable builds. Model weights are published by Stability AI through its official channels.