Stable Diffusion

Name: Stable Diffusion
Author: Stability AI

by Stability AI4.3

Stable Diffusion is the umbrella name for Stability AI's open image models. Because it runs on your own machine or server, it is the answer whenever data must not go to an external service.

Commercial use OK
Free tier

Try Stable Diffusion on the official siteAPI / OSS

Screenshot of the Stable Diffusion website

Pricing: API / OSS
Free tier: Open source
Vendor: Stability AI
Commercial use: Permitted
Japanese UI: Not confirmed
Enterprise contract: Not confirmed

What it is good for

Creative teams with engineering capability
Corporate R&D that cannot send data outside
Fine-tuning and heavily controlled generation

Strengths and trade-offs

Strengths

Runs locally or on your own servers
Weights are free; an API is also available
LoRA, Textual Inversion and ControlNet give deep control

Trade-offs

Setup needs real engineering effort — GPU, ComfyUI or A1111
Operating cost is far higher than a cloud SaaS

Editors' review

The open-source standard — local execution and full control

Stable Diffusion covers Stability AI's open image models. Because they run on a local machine or your own server, they fit requirements where images and prompts must not leave the building.

Weights are free to download and an API is available on usage-based pricing. The ecosystem around it — LoRA and Textual Inversion for fine-tuning, ControlNet for structural control — has led the open-source field for years.

Setting it up takes engineering effort: a GPU environment and an interface such as ComfyUI or A1111. Compared with a cloud service, the running cost in time is significant.

Verdict

For technically capable creative teams and for corporate R&D that cannot send data out. If you want convenience, use Midjourney or Adobe Firefly.

Compiled from official documentation and public sources, with our own editorial judgement. We do not publish figures we cannot source.