Stable Diffusion 3.5 launches with improved prompt adherence and diverse image outputs

Stable Diffusion 3.5 was released on October 22, featuring improved prompt adherence and diverse outputs. It includes three customizable models designed for various uses, compatible with consumer hardware, and free for startups with annual revenue under $1 million. This update addresses issues from the previous version, which faced criticism for poor image quality and prompt adherence. Stability AI claims the new models are more robust and can generate images in various styles, including 3D art. Initial user feedback has been mixed, with comparisons to other image generators. Stability AI asserts that the new models produce more diverse outputs, representing a wider range of skin tones and features without extensive prompting.