Stability AI releases new Stable Diffusion 3.5 models
Stability AI has announced the release of Stable Diffusion 3.5, a new generation of image generation models. The launch includes several model variants for different user groups.
Two models are immediately available: Stable Diffusion 3.5 Large and Stable Diffusion 3.5 Large Turbo. A third variant, Stable Diffusion 3.5 Medium, is set to release on October 29. These models can run on consumer hardware and are free to use for most applications under the Stability AI Community License.
The Stable Diffusion 3.5 Large model has 8 billion parameters and is designed for a resolution of 1 megapixel. Stable Diffusion 3.5 Large Turbo is a faster version that generates images in just 4 steps.
The upcoming Stable Diffusion 3.5 Medium model, with 2.5 billion parameters, is intended to run on consumer hardware and generate images with a resolution of 0.25 to 2 megapixels.
Ad
Stable Diffusion should be competitive again
Stability AI claims that Stable Diffusion 3.5 Large leads the market in prompt adherence and can compete with larger models in terms of image quality. Stable Diffusion 3.5 Large Turbo aims to offer fast inference times with competitive image quality and prompt adherence. Most recently, the company disappointed its community with Stable Diffusion 3 Medium. The community increasingly turned to new models such as FLUX.
The new models are released under Stability AI’s Community License, which allows free use for non-commercial purposes and commercial use for companies with annual revenue under $1 million.
Model weights are available on Hugging Face for self-hosting. Additionally, there are partnerships with platforms such as the Stability AI API, Replicate, ComfyUI, and DeepInfra.
