Stable Diffusion 3.5
A family of image generation models from Stability AI that runs on consumer-grade hardware and is free for both commercial and non-commercial use within the terms of the Stability AI Community License.
About Stable Diffusion 3.5
Stable Diffusion 3.5 is a family of image generation models from Stability AI, available in multiple variants: the Large version has 8.1 billion parameters, making it the most capable base model in the Stable Diffusion family, with strong generation quality and prompt adherence, suited to professional use cases at 1 megapixel resolution. Each model is highly optimized for its size and can run on consumer-grade hardware. Users can download all Stable Diffusion 3.5 models from Hugging Face and obtain the inference code from GitHub. Regarding terms of use, under the Stability AI Community License, individuals and organizations may use the models free of charge for non-commercial purposes (including scientific research); startups, small and medium-sized businesses, and creators with annual gross revenue below 1 million USD may also use the models commercially for free; and users retain ownership of the media content they generate, free from restrictive licensing terms.
Best for
Best use cases
Research institutions or individuals conducting image generation research
Startups, small and medium-sized businesses, and creators with annual revenue below 1 million USD carrying out commercial creative work
Running image generation models locally on consumer-grade hardware
Features
Multi-size model lineup
Offers model versions in different sizes, including Large, which has 8.1 billion parameters and is the most capable base model in the Stable Diffusion family, delivering high generation quality and prompt adherence, suited to professional scenarios at 1 megapixel resolution.
Runs on consumer-grade hardware
Each model is highly optimized for its size and can run on consumer-grade hardware.
Free use under the community license
Under the Stability AI Community License, individuals and organizations may use it free of charge for non-commercial purposes (including scientific research); startups, small and medium-sized businesses, and creators with annual revenue below 1 million USD may use it commercially for free; and ownership of generated content is retained by the user.
Publicly downloadable
All Stable Diffusion 3.5 models can be downloaded from Hugging Face, and the inference code is available on GitHub.
Pros
- Free for non-commercial use (including scientific research)
- Free for commercial use with annual gross revenue below 1 million USD
- Runs on consumer-grade hardware
- Users retain ownership of generated content
- Models downloadable from Hugging Face, with inference code released on GitHub