Stability AI has been at the center of the text-to-image revolution with the release of the Stable Diffusion family of models. In recent months, we have seen an explosion of breakthroughs in research areas related to text-to-image models, such as 3D classification, controllable image editing, image personalization, synthetic data augmentation, graphical user interface prototyping and several others.

Incorporating some of those breakthroughs led Stability to gradually improve Stable Diffusion. The latest result of this work was the release of SDXL, a very advanced latent diffusion model designed for text-to-image synthesis. With this release, SDXL is now the state-of-the-art text-to-image generation model from Stability AI.

SDXL is now available via ClipDrop, GitHub or the Stability AI Platform.

Let's dive into the details...