Muse Video is the upcoming AI video generator from Meta Superintelligence Labs (MSL), previewed on July 7, 2026 alongside the launch of Muse Image. It is the lab's first video model — and, unusually for a Meta announcement, you cannot try it anywhere yet.
What Meta has confirmed
- Shared foundation with Muse Image. Muse Video is built on the same pretraining base as Muse Image. Meta trained image and video capability as one family, which typically means consistent visual style and editing behaviour across both.
- Native audio. Muse Video generates sound together with the picture — dialogue, ambience, effects. That puts it in the Sora 2 / Veo 3 generation of video models, not the silent-clip era.
- "Exceptional visual fidelity" is Meta's headline pitch, consistent with how Muse Image debuted at No. 2 on the Arena human-preference leaderboard.
- Rollout target: "coming soon to creators and Meta AI" — which reads like an Instagram/Reels-first strategy rather than a standalone product.
What Meta has not said
- A release date — "coming soon" is the only timing language used so far.
- Regions. Muse Image launched US-only; expect the same pattern.
- An API. Muse Image shipped without one; there is no indication Muse Video will differ.
- Clip length, resolution, watermarking, usage limits or pricing.
How it relates to the rest of the Muse family
The Muse family currently has three members: Muse Image (live, US, inside Meta products), Muse Video (previewed), and Muse Spark 1.1 (the language model, live with a developer API). The family launched over three days in July 2026 as the first public output of Meta Superintelligence Labs under Alexandr Wang.
While you wait
If you need AI video today, the realistic options are Sora 2 and Veo 3 — see our side-by-side comparison. If what drew you here is the Muse aesthetic on the image side, the top-ranked alternatives are usable from any country — our how-to-try guide walks through them.