Meta's first in-house AI video model — text-to-video with native audio, previewed alongside Muse Image and coming soon to creators and Meta AI.
Last verified · 2026-07-08 · by Moe Ameen
Muse Video is Meta's first in-house AI video generation model, previewed on July 7, 2026 alongside Muse Image. Both are the first media-generation models from Meta Superintelligence Labs (MSL), the group led by Alexandr Wang. Muse Video is built on the same pretraining base as Muse Image, and its headline capability is native audio — it generates the soundtrack along with the picture from a single text prompt, rather than producing a silent clip you score afterward.
Unlike Muse Image, which shipped to real users the same day, Muse Video launched as an early preview, not a public product. Meta showed sample generations to demonstrate where its video quality stands and said the model is "coming soon to creators and Meta AI," without a firm release date. At preview, Meta reported Muse Video ranking around #3 on the Arena human-preference leaderboard for text-to-video, trailing the leading entries from OpenAI and Google.
Meta was candid about the gaps. It flagged two areas still being refined: audio-video synchronization, and physically accurate fast motion — the same weak spots most current text-to-video models share. No resolution, clip length, or pricing has been published; those details will land when the model opens to creators.
The honest framing: Muse Video is an early-preview generator, not a content-production tool, and not yet something you can log in and use. When it opens, it will make short clips with sound inside Meta's own apps — it won't caption those clips for other feeds, keep a brand voice across a posting week, cut a long video into shorts, build carousels or blogs, or schedule and publish anywhere outside Meta's surfaces. Treat every spec here as a preview-window snapshot; confirm against Meta's own announcement before relying on it.
Muse Video's native-audio trick is the interesting part — one prompt yields a clip that already has sound — but the output is a single clip that, when the model opens, will live inside Meta AI or a Meta app. That is one moment, in one place, in one aspect ratio, with whatever audio the model generated. A creator needs that moment in ten places. Kompozy is the layer that gets it there: bring a Muse Video clip in and it becomes a Clipped Short cut for feeds, with your own branded captions burned in (auto-transcribed, so they track the native audio), the frame reframed to 9:16, 1:1, or 4:5 per destination, and hook text stacked through HyperFrames so the muted first second still reads on autoplay.
From that one clip Kompozy fans a full unit across five buckets: the short for TikTok, Reels, Shorts, and YouTube; native Text Posts and a thread for X and LinkedIn; a Quote Graphic or brand-exact Carousel pulling the clip's best line; and a Blog Article and Email Newsletter around the idea — every piece held to one voice by your Persona Brief and banned-word filters, so the volume reads as your brand instead of raw model output. Then Kompozy closes the loop Muse leaves open: it schedules and publishes the whole set across all nine social platforms plus blog and email from a single queue, with Autopilot and a per-post review pipeline. And where Muse's clip is a one-off, Kompozy generates the recurring video Meta can't — HeyGen-driven Persona Shorts and Persona Frames built on a face-locked identity you own, so your channel has a consistent presenter every week, not a different generated face each time.
Muse Video is Meta's first in-house AI video generation model, previewed July 7, 2026 by Meta Superintelligence Labs alongside Muse Image. It generates short video from a text prompt with native audio — the soundtrack is produced together with the picture. It was shown as an early preview and is 'coming soon to creators and Meta AI' rather than publicly available at launch.
Not as a public product. Meta previewed it on July 7, 2026 to show its video quality and said access is 'coming soon to creators and Meta AI,' without a firm release date. Muse Image launched to users the same day, but Muse Video did not.
Yes. Native audio is its headline feature — Muse Video generates a soundtrack alongside the picture from a single prompt, instead of producing a silent clip you score later. Meta has said audio-video synchronization is one of the areas it is still refining.
At preview, Meta reported Muse Video ranking around #3 on the Arena human-preference leaderboard for text-to-video, behind the leading models from OpenAI and Google. Meta also flagged remaining gaps in audio-video sync and physically accurate fast motion. No resolution, clip length, or pricing has been published.
When the model opens, generate the clip in Meta AI, then bring it into Kompozy. Kompozy auto-captions it, reframes it per platform, and fans it into a Clipped Short, quote graphics, a carousel, text posts, a blog, and a newsletter in your voice via the Persona Brief — then schedules and publishes the set across nine platforms plus blog and email.