Stability AIStable Diffusion

Stable Diffusion 3.5 Large

Stable Diffusion 3.5 Large is a text-to-image diffusion transformer with downloadable weights and hosted API access. Its architecture combines three text encoders and query-key normalization.

Model design and access

The official card includes ComfyUI and Diffusers workflows, plus a low-bit loading example. Quantization changes memory use and inference behavior; it is a deployment choice within this release. The model is designed for image creation, including text within images, rather than factual reconstruction of a scene.Stable Diffusion 3.5 Large official model documentation

Stable Image Ultra is a hosted image service using models that include SD 3.5. Its API supports an optional starting image and strength control, and returns approximately one-megapixel images. Ultra is grouped here as a related service, not a separate downloadable checkpoint.Stable Image Ultra API reference

Practical use

Practical guidance. Large is suited to workflows that need control over local generation, adapters or image pipelines. Include the text encoders when assessing memory needs. For layouts with required words, review spelling and placement at full size; a model that can render text still needs checks before publication.

Specifications & access

Stable Diffusion 3.5 Large specifications
DetailValue and conditions
AccessAPI, Downloadable
Model identifierssd3.5-largeStable Image Ultra
Model size8 billion parameters.Stable Diffusion 3.5 Large official model documentation
LicenseStability AI Community License.Stable Diffusion 3.5 Large official model documentation
Text encodersCLIP-L, OpenCLIP-G and T5-XXL.Stable Diffusion 3.5 Large official model documentation

Related models