Stable Diffusion 3.5 Medium

Stable Diffusion 3.5 Medium is a text-to-image generative model developed by Stability AI. Its model card identifies the architecture as MMDiT-X, an improved Multimodal Diffusion Transformer that conditions image generation on text. This makes it a different architecture generation rather than a minor update to Stable Diffusion 1.4. ‘Medium’ names the member of the 3.5 family; it does not describe image quality or hardware size. Its role is image synthesis, not visual question answering. TT-Metal’s matrix lists 512x512 images, batch 1, on a single n150 Wormhole device. Only that Wormhole row appears in the snapshot, unlike Stable Diffusion 1.4, which has both Wormhole and Blackhole alternatives.