Meta’s Superintelligence Labs has introduced Muse Image for image generation and a preview of Muse Video, both built on the Muse architecture. Muse Image pairs with the Muse Spark language model to support editing, sketching, annotations, text rendering, and real-time search-driven reference blending—users can even @mention a public account to replicate its visual style. The early Muse Video build generates video and audio in a single stream with precise synchronization, and all outputs are protected by a Content Seal watermark resistant to compression and cropping. Muse Image is available for free with daily limits, with a paid pro tier, while the video generator is expected to launch in the coming months.

Meta: Introducing Muse Image, Muse Video, and Muse Spark Language Model