Open generative artificial intelligence startup Stability AI Ltd. has released the most advanced version of its flagship text-to-image model, Stable Diffusion XL 1.0.

Launched today, SDXL 1.0 will be made available through a range of platforms, including the Stability AI application programming interface, Stability AI’s GitHub page and its Clipdrop and DreamStudio consumer applications. It will also be featured on Amazon Bedrock, a fully managed service on Amazon Web Services Inc.’s cloud platform that provides access to a selection of foundational AI models.

Fittingly, SDXL 1.0 was announced at the annual AWS Summit New York, and Stability AI said it’s further acknowledgment of Amazon’s commitment to providing its customers with access to the most powerful generative AI models around.

Stability AI Chief Executive Emad Mostaque said SDXL 1.0 represents the next step of the company’s innovation heritage. “Unveiling 1.0 on Amazon Bedrock demonstrates our strong commitment to work alongside AWS to provide the best solutions for developers and our clients,” he explained.

An earlier release of SDXL appeared in April. Unlike generative AI models such as ChatGPT, which is focused on creating text based on users’ prompts, SDXL is all about image generation. It’s said to bring next-level photorealism to AI-powered image generation, with enhanced capabilities around image composition and, in particular, face generation (pictured). In addition, developers can use much simpler prompts to create descriptive images, the company says.

SDXL also supports “image-to-image” prompts, meaning users can input one image and then add some text prompts to create more detailed variations of that picture. The inpainting feature makes it simple to reconstruct missing parts of an image too, and the outpainting feature allows users to extend existing images.

Stability AI said SDXL 1.0 has been significantly improved since the earlier release. Now, it’s said to generate even more vibrant and accurate colors, with superior contrast, lighting and shadows. In addition, users can now yield full 1-megapixel resolution images in multiple aspect ratios, the company said.

Perhaps the best things about SDXL 1.0 are its open-source nature and ability to run on relatively simple hardware – two capabilities that make it much more accessible than competing image generation models.

“The addition of Stability AI’s SDXL 1.0 to Amazon Bedrock expands the foundation models available to customers to now include the leading text-to-image generator on the market,” said Swami Sivasubramanian, AWS’ vice president of database, analytics and machine learning.

SDXL 1.0 is just the latest addition to Stability AI’s growing library of AI models. In April, it announced the release of StableLM, which more closely resembles ChatGPT with its ability to generate text and code. Last month, it debuted Stable Doodle, a “sketch-to-image” tool that works by converting simple sketches into more illuminating images.

Image: SDXL/Stability AI