Stable Diffusion XL 1.0 text-to-image model launches with built-in fine-tuning feature

Stability AI's most advanced text-to-image model is now available via API, Github, and Amazon Bedrock, bringing cutting-edge AI image technology to developers and customers.

Stability AI announced the release of Stable Diffusion XL (SDXL) 1.0, the latest and most advanced text-to-image model in its suite. The model is available on Amazon Bedrock, a fully managed service from Amazon Web Services (AWS) that provides API access to foundational models from leading AI companies and Amazon Titan.

The model is accessible through the Stability AI API, GitHub page, and its Clipdrop and DreamStudio consumer applications. It is also available on Amazon SageMaker JumpStart, an ML hub that provides access to algorithms, models, and ML solutions.

SDXL 1.0 follows the research-only release of SDXL 0.9 in June. The model has been refined to produce more vibrant and accurate colors, improved contrast, lighting, and shadows, the company says. In addition, the imaging process has been streamlined to deliver faster output with full 1-megapixel resolution images (1024x1024) in multiple aspect ratios.

SDXL comes with an integrated Dreambooth feature

A new fine-tuning beta feature is also being introduced that uses a small set of images to fine-tune SDXL 1.0, allowing users to specialize the generation to specific people or products using as few as five images.

SDXL 1.0 is one of the most powerful open-access image models available, with a 3.5B parameter base model and a 6.6B parameter model ensemble pipeline. You can try out the model for free using Stability.ai's Clipdrop.

In terms of quality, SDXL is probably the only commercially available image model that comes close to Midjourney's level of quality, and thanks to its open-source nature, it offers even more features and options - although it is a bit more difficult to use if you want to take full advantage.

Since the SDXL beta launched in April, ClipDrop users have generated more than 35 million images using the model, and Stability AI's Discord community has generated an average of 20,000 images per day, according to the company.

Stability AI has a strong relationship with AWS, having selected it as its preferred cloud provider in 2022 and building its foundational models on Amazon SageMaker. Stability AI's Stable Diffusion model was one of the first to be made available on Amazon Bedrock when it launched in April.