Meta introduced the image generation model Muse Image and the earlier preview Muse Video, developed by Meta Superintelligence Labs. Muse Image is already available in the Meta AI application, on meta ai, in Instagram Stories (USA) and WhatsApp (restricted countries). The model is positioned as the most advanced in the Meta portfolio: it follows more precisely, supports editing, image compilation from multiple referrals and uses social context Instagram.

Muse Image works in an agent mode: it uses web search, tools to write and execute code, and iteratively completes the image during generation. Integration with Muse Spark allows you to create animated GIFs, embedded image sites and interactive visual games.

Muse Video is built on the same foundation of pre-training, providing high visual quality and native audio support. Meta notes that it continues to work on syncing audio-video and physically accurate transmission of fast motion. All images created through Muse Image in the Meta AI application receive an invisible water mark, the Content Seal, which is retained after framing, compression or screencast. According to Arena’s metrics, Muse Image ranks second in the text-to-image and image editing categories, behind only GPT Image 2.