Stable Diffusion Releases 3.5 Image Model

Stability AI has released Stable Diffusion 3.5, introducing significant improvements in image generation capabilities.

The new release includes multiple variants designed to run on standard consumer hardware, with the flagship Large model featuring 8 billion parameters and operating at 1 megapixel resolution.

The Breakdown

• The release includes two immediate variants: Stable Diffusion 3.5 Large for professional-grade image generation, and Large Turbo, a faster version that generates images in just 4 steps.

• A Medium version will launch on October 29th, followed by ControlNets for advanced control features

• The models excel at prompt adherence and can generate diverse visual styles including 3D, photography, painting, and line art

• The licensing structure makes it accessible for most users - free for non-commercial use and commercial applications for businesses with under $1M in annual revenue

Industry Impact

This release represents a significant step forward for content creators and production professionals. The ability to run these models on standard hardware, combined with their enhanced customization capabilities, makes advanced AI image generation more accessible to production teams of all sizes.

The improved prompt adherence and diverse style capabilities could streamline pre-visualization workflows and concept art development, while the permissive licensing structure enables broad adoption across the creative industry.