Stable Diffusion 3.5 - A highly customizable, open-source text-to-image model series by Stability AI.

## Overview of Stable Diffusion 3.5 Stable Diffusion 3.5 is an open-source text-to-image model series developed by Stability AI. It includes multiple variants such as Stable Diffusion 3.5 Large and Stable Diffusion 3.5 Medium. These models are designed to be highly customizable, run efficiently on consumer-grade hardware, and are available for free under the Stability AI Community License for both commercial and non-commercial use, provided the annual revenue is below $1 million. ## Key Features of Stable Diffusion 3.5 The key features of Stable Diffusion 3.5 include: - Multiple model variants: Stable Diffusion 3.5 Large (8.1 billion parameters) and Stable Diffusion 3.5 Medium (2.5 billion parameters). - Resolution capabilities: Large supports 1 megapixel, while Medium supports 0.25–2 megapixels. - Hardware requirements: Medium requires 9.9 GB VRAM (excluding text encoder) for optimal performance. - High image quality and strong prompt adherence. - Support for fine-tuning and LoRA for customization. - Efficient operation on consumer-grade hardware. - Diverse output styles, including 3D, photography, and painting. ## Accessing Stable Diffusion 3.5 Users can access Stable Diffusion 3.5 by: - Downloading the model weights from [Hugging Face](https://huggingface.co/stabilityai). - Obtaining the inference code from [GitHub](https://github.com/Stability-AI/sd3.5). - For some models like 3.5 Large, access is also available through [Amazon SageMaker JumpStart](https://stability.ai/news/stable-diffusion-3-5-large-is-now-available-in-amazon-sagemaker-jumpstart). ## Licensing Terms for Stable Diffusion 3.5 Stable Diffusion 3.5 is available under the Stability AI Community License, which allows: - Free use for research and non-commercial projects. - Free use for commercial projects with annual revenue below $1 million. Organizations with annual revenue above $1 million must contact [Stability AI Enterprise](https://stability.ai/enterprise) for licensing. Users retain ownership of the images generated. ## Technical Specifications of Stable Diffusion 3.5 Models The technical specifications of Stable Diffusion 3.5 models are as follows: - **Stable Diffusion 3.5 Large**: 8.1 billion parameters, supports 1 megapixel resolution, optimized for professional use. - **Stable Diffusion 3.5 Large Turbo**: Unspecified parameters and resolution, capable of generating high-quality images in 4 steps with fast inference speed. - **Stable Diffusion 3.5 Medium**: 2.5 billion parameters, supports 0.25–2 megapixel resolution, requires 9.9 GB VRAM (excluding text encoder) for optimal performance, balanced for quality and customization on consumer-grade hardware. ## Applications of Stable Diffusion 3.5 Stable Diffusion 3.5 can be applied in various scenarios: - **Stable Diffusion 3.5 Large**: Suitable for professional use in media, entertainment, and retail content creation. - **Stable Diffusion 3.5 Large Turbo**: Ideal for applications requiring quick image generation. - **Stable Diffusion 3.5 Medium**: Best for small to medium-sized projects on consumer-grade hardware, balancing quality and customization. The models also support fine-tuning and optimization for specific use cases. ### Citation sources: - [Stable Diffusion 3.5](https://stability.ai/news/introducing-stable-diffusion-3-5) - Official URL Updated: 2025-03-26