OpenWeave releases Muse-XL: a 1-trillion parameter multimodal model optimized for design workflows

AI · 5 min read

OpenWeave today announced Muse-XL, a 1-trillion parameter multimodal model that blends text, images, and layout understanding to accelerate end-to-end design work. Unlike consumer-facing image generators, Muse-XL is tuned to output not only raster images but also componentized SVGs, CSS snippets, and accessible markup that designers can drop into prototypes.

The company included a "trace" mode that highlights which pieces of the prompt, wireframe, or brand tokens influenced specific parts of the output—helping teams audit decisions and iterate faster. Muse-XL also supports deterministic sampling, reducing surprising variations during collaborative sessions.

OpenWeave is shipping Muse-XL with a plugin SDK for Figma, Framer, and in-house tools and supports hybrid deployment so organizations can run inference on-prem or in trusted clouds. Early partner tests claim Muse-XL reduces mockup iteration time by roughly 30–45% on typical landing-page tasks.