Stable Diffusion

Stable Diffusion is Stability AI's family of image generation models. The current line includes Stable Diffusion 3.5 in Large, Turbo, and Medium sizes, alongside Stable Diffusion XL and SDXL Turbo. Models can be run on your own hardware under a self-hosted license, called through the Stability AI API, used through cloud partners, or tried in the browser-based Stability Assistant. Stability AI presents the models as open, adaptable technology for creators, developers, and enterprises.

Key Features

• Stable Diffusion 3.5 Large, a one-megapixel model aimed at professional work, and Turbo, which generates in four steps.
• Stable Diffusion 3.5 Medium, designed to run on consumer hardware, plus Stable Diffusion XL and SDXL Turbo.
• Editing services that erase objects, inpaint, outpaint, remove backgrounds, search and recolor, and replace backgrounds with relighting.
• Upscaling in Creative, Conservative, and Fast modes.
• Control tools for sketch, structure, and style guidance.
• Deployment through a self-hosted license, the Stability AI API, cloud partners, or the Stability Assistant web app.
• A Community License that covers free use for individuals and organizations under one million US dollars in annual revenue.

Who is it for

• Researchers who need models they can inspect, fine-tune, and run locally.
• Developers building image generation into an application through an API.
• Small studios and independent creators working without a large budget.
• Enterprises that need a commercial license, support, and custom model training.
• Teams that must keep image data inside their own infrastructure.

What can you do with Stable Diffusion

• Generate images from text prompts in a wide range of styles, from photography to line art.
• Edit an existing image by erasing objects, inpainting, or extending the frame.
• Remove or replace a background and relight the subject to match.
• Upscale a low-resolution image with a creative or conservative pass.
• Guide composition with sketch, structure, or style controls.
• Self-host a model, call it through the API, or use it in a cloud partner environment.

How does Stable Diffusion work

• Choose a model size and a deployment route: self-hosted, API, cloud partner, or the Assistant app.
• Obtain a license or API key, then load the model in your own environment or send a request.
• Write a prompt and set generation options such as resolution and style.
• Review the output and refine it with editing, control, or upscaling services.
• Integrate the calls into your product, or export the finished image.
• Move to an enterprise license when usage or revenue crosses the community thresholds.

Pricing

• The Community License is free for individuals, researchers, and organizations earning under one million US dollars per year.
• It covers the Stable Diffusion 3.5 suite, SDXL Turbo, Stable Audio 3.0, and Stable Fast 3D.
• An Enterprise License with commercial use, implementation support, and custom training is sold at custom pricing.
• API usage is credit-based, with one credit equal to 0.01 US dollars; Stable Image Ultra costs 8 credits per image.

FAQ

Is Stable Diffusion free?

The Community License is free for individuals and for organizations with less than one million US dollars in annual revenue. It permits research, non-commercial, and commercial use within those limits.

What happens above one million dollars in revenue?

Organizations above that threshold need an Enterprise License, which adds commercial use, implementation support, and options such as custom model training and consulting.

Can I run the models on my own hardware?

Yes. A self-hosted license allows deployment in your own environment, which gives control over data and customization. Stable Diffusion 3.5 Medium is specifically designed to run on consumer hardware.

Does Stability AI offer an API?

Yes. The Stability AI API exposes image generation, editing, upscaling, and control services on a credit basis, with one credit equal to 0.01 US dollars.

Which models are included in the Community License?

The license lists the Stable Diffusion 3.5 suite, SDXL Turbo, Stable Audio 3.0, and Stable Fast 3D. Stable Diffusion XL is also available as an additional image model.