Stable Diffusion
ActiveStability AI (open weights) · founded 2022
Open-weight models you can self-host for full control and privacy.
Stable Diffusion is a family of open-weight image models you can run locally or self-host (often via ComfyUI or Automatic1111). It offers unmatched customization, fine-tuning, and privacy, at the cost of a much steeper setup and hardware requirements.
At a glance
- Category
- AI Image Generation
- Primary use case
- Self-hosted, customizable image generation
- Entry price
- Free
- Free plan
- Free (self-hosted)
- API
- Yes
- Open source
- Yes
- Evidence level
- Research / specifications only
- Last checked
- 20 Aug 2026
Pricing
Last checked 20 Aug 2026
| Free plan | Free (self-hosted) |
|---|---|
| Trial | None |
| API pricing | Stability AI API usage-based |
| Usage limits | Limited by your hardware (self-host) or API credits |
Model weights free/open; running locally costs hardware, API costs credits.
What it does
Secondary use cases
- Fine-tuning
- Workflows (ComfyUI)
- Private generation
Input types
Text, Image
Output types
Image
Models used
SD 3.5, SDXL, Flux (community)
Platforms
Integrations
Privacy & commercial rights
Data retention
Full local control if self-hosted
Trains on your data
No (per stated policy)
Enterprise privacy
Not disclosed
Self-hosting / local
Available
SSO
Not disclosed
Compliance
Not disclosed
Commercial use
Permitted — Open-weight licenses permit commercial use (check specific model license)
Open source
- License
- Mixed (open-weight licenses)
- Last commit
- active
- Self-host requirements
- Dedicated GPU recommended (8GB+ VRAM)
- GPU required
- Yes
Pros & cons
Pros
- Full control and privacy
- Unlimited customization/fine-tuning
- No per-image cost when self-hosted
Cons / trade-offs
- Steep setup
- Requires capable GPU
- No polished consumer UX
Alternatives to Stable Diffusion
Considering a switch, or comparing before you buy? See the full alternatives guide.