
Stable Diffusion is Stability AI’s family of image generation models. Version 3.5 comes in Large, Large Turbo and Medium, and runs on your own machine, through the API or in the Stable Assistant web app.

Terms checked on stability.ai in November 2026.
Stable Diffusion made open-source image generation mainstream. Unlike Midjourney or Nano Banana, you can download the models and run them on your own computer or server, without sending your images to a third party.
The current version, Stable Diffusion 3.5, comes in three sizes:
Around the model, Stability AI offers editing tools through its API: erase objects, inpaint and outpaint, upscale, or control composition from a sketch.
Stable Diffusion suits teams that want control: training a house style, producing at volume without a subscription, or handling confidential visuals locally. For brand campaigns, we mostly use it to explore directions and produce variations, then take the art direction back by hand. If you just want great images without installing anything, Krea or Nano Banana will be faster.
Is Stable Diffusion free?
Yes, for personal use and for businesses under $1 million in annual revenue, under Stability AI’s Community licence. Above that, you need an Enterprise licence.
Can I use Stable Diffusion without installing it?
Yes, through the Stable Assistant web app or the many services that embed the model.
Which version should I pick?
Large for quality, Large Turbo for speed, Medium if your hardware is modest.