A.I.
Meta · Muse
Muse Video
Muse Video is Meta’s video generation model in the Muse family (same pretraining base language as Muse Image on the introduce blog). Meta’s public language has been preview or coming soon for creators and Meta AI.
- Status
- Preview, coming soon
- Last checked
- 12 Oct 2026newest source date on this page
What it is
Meta previewed Muse Video on 7 July 2026, alongside the launch of Muse Image (Meta, 7 Jul 2026). It is built on the same pretraining base as Muse Image, meaning both start from the same first round of training, and it makes video with sound generated at the same time.
Status
As of Meta's post, Muse Video is coming soon to creators and to Meta AI. It is not yet generally available, and Meta has not published pricing or an API.
What Meta says
- Competitive on following the prompt, image quality, and consistency from frame to frame.
- Still weak on keeping sound in sync with video and on fast motion that obeys physics. Meta says it is working on both.
- No. 3 on Arena's human-preference ranking for text-to-video as of 5 July 2026, per Meta. Arena is a public site where people vote between two models' outputs.