Stability AI, a leading open generative AI company, announced the release of Stable Diffusion XL (SDXL) 1.0, the latest and most advanced of its flagship text-to-image suite of models.
SDXL 1.0 will be featured on Amazon Bedrock, the fully managed service from Amazon Web Services Inc. (AWS) that provides access to foundation models from leading AI startups and Amazon Titan via an API. The announcement is made at the annual AWS Summit New York, and is further acknowledgment of Amazon‘s commitment to bringing its customers the most powerful text-to-image foundation models like SDXL 1.0 from Stability AI.
“The latest SDXL model represents the next step in Stability AI‘s innovation heritage and ability to bring the most cutting-edge open access models to market for the AI community,” said Emad Mostaque, Chief Executive Officer of Stability AI. “Unveiling 1.0 on Amazon Bedrock demonstrates our strong commitment to work alongside AWS to provide the best solutions for developers and our clients.”
CIO INFLUENCE: CIO Influence Interview with Pete Lilley, Vice President and GM at Instaclustr
Starting today, the model is accessible via the Stability AI API, Stability AI‘s Github page and its Clipdrop and DreamStudio consumer applications. Alongside its release on Amazon Bedrock, SDXL 1.0 is available on Amazon SageMaker JumpStart, a machine learning (ML) hub that provides access to algorithms, models, and ML solutions so customers can quickly get started with ML.
This follows the research-only release of SDXL 0.9 in June, which provided a significant number of use case testing data that allowed developers to fine-tune the model for better results.
The refining process has produced a model that generates more vibrant and accurate colors, with better contrast, lighting, and shadows than its predecessor. The imaging process is also streamlined to deliver quicker results, yielding full 1-megapixel (1024×1024) resolution images in seconds in multiple aspect ratios.
Stability API’s new fine-tuning beta feature will use a small set of images to fine-tune SDXL 1.0. With as few as five images, users will be able to specialize generation on specific people, products, and more. This feature is currently in limited testing with a number of early access partners, and will be released in the coming weeks.
CIO INFLUENCE: JFrog Software Supply Chain Platform Delivers 393% ROI According to Total Economic Impact Study
SDXL 1.0 has one of the largest parameter counts of any open access image model, boasting a 3.5B parameter base model and a 6.6B parameter model ensemble pipeline. The refiner adds more accurate color, higher contrast, and finer details to the output of the base model.
Since the SDXL beta launch on April 13, ClipDrop users have generated more than 35 million images using the model, while Stability AI‘s Discord community has generated an average of 20,000 images per day.
“Model choice is paramount to maximize the value customers get from generative AI,” said Swami Sivasubramanian, Vice President of Database, Analytics, and Machine Learning at AWS. “That’s why we built Amazon Bedrock to give customers access to a wide range of the top-performing foundation models, allowing them to find and customize the right foundation model to fit their needs with a fully managed experience. By expanding Amazon Bedrock‘s selection with the addition of Stability AI‘s SDXL 1.0 model, we’re giving customers access to a state-of-the-art text-to-image model to build and scale exciting, new generative AI applications.”
In 2022, Stability AI selected AWS as its preferred cloud provider, building its foundation models on Amazon SageMaker.When Amazon Bedrock launched in April of this year, Stability AI‘s Stable Diffusion was one of the first foundation models to be made available on Amazon Bedrock. Today’s release of SDXL 1.0 on Amazon Bedrock expands Stability AI‘s and AWS‘s relationship and commitment to helping customers build and scale generative AI applications easier and quicker.
CIO INFLUENCE: World Password Day: Password advice for CIOs
[To share your insights with us, please write to sghosh@martechseries.com]
