Stable Diffusion 3

The most capable open-source text-to-image model

Stable Diffusion 3 employs an innovative diffusion transformer technique enhanced with flow matching and various other improvements. Its ability to understand prompts has significantly improved compared to previous models.