Background
Muse Video is Meta Superintelligence Labs' first generative video model, previewed on 7 July 2026 alongside the image model Muse Image. It generates video with native audio support, composing sound effects, ambience and dialogue in the same generation pass as the visuals, and adds physics grounding, reusable character identity and clip remixing on top. It ranks third on the Arena text-to-video leaderboard, though Meta has flagged open gaps in audio-video synchronisation and physically accurate fast motion.
The model is preview-only, with no public resolution, duration or pricing confirmed. Meta positions it against rivals such as OpenAI's Sora and Runway as it builds out in-house generative tooling for Instagram, WhatsApp and Facebook.
What is Muse Video?
When was Muse Video announced?
Written with AI from Lowdown's published reporting. Not individually reviewed by an editor. Updated 15 July 2026. Editorial standards.