Stability AI launches advanced open-source Text-to-Image Generative Model, Stable Diffusion XL 1.0

Stability AI has introduced Stable Diffusion XL 1.0, their new open-source text-to-image generative AI model with improved color vibrancy, accuracy, contrast, shadows, and lighting. The model, which is viewed as the company's most advanced, has 3.5 billion parameters, enabling it to generate 1-megapixel resolution images in seconds and support multiple aspect ratios. Unlike its predecessor, Stable Diffusion XL 1.0 is customizable, providing users with options to fine-tune different concepts and styles.

The model exhibits advanced text generation capabilities, producing clear and accurate logos, text, and fonts in its generated images. However, the open-source version of Stable Diffusion XL 1.0 could potentially be misused by malicious actors to produce harmful content, including nonconsensual deepfakes. To counter this, Stability AI has implemented measures to filter unsafe imagery and block problematic terms in the tool, as the company is committed to continually improving the safety features of Stable Diffusion and reducing the generation of harmful content.

As of today, the latest model, Stable Diffusion XL 1.0 can be accessed through various channels, including the Stability AI API, Stability AI's Github page, and its consumer applications, Clipdrop and DreamStudio. Additionally, it is now available on Amazon SageMaker JumpStart, an ML hub offering algorithms, models, and ML solutions, enabling customers to swiftly embark on their machine learning journey.