Correction, 2026-10-01: the table listed Kling v3 Standard for cheap variants without saying that on 8frame Kling v3 is image-to-video only and needs a start image. It now says so and gives text-only options: Kling 2.6 Pro, or Wan 3.0 at 480p.
Meta Muse Video is the first video generation model from Meta Superintelligence Labs, previewed on July 7, 2026 alongside Muse Image. Meta says it is "built on the same pretraining base" as Muse Image and "delivers exceptional visual fidelity with native audio support", and that it "ranks No. 3 in human-preference Elo for text-to-video" on Arena at the time of writing. Meta's own words on availability: "coming soon to creators and in Meta AI." As of our check on 2026-09-30, there is no public API, resolution, length or pricing.
Correction, 2026-09-30: the first version of this page said Muse Video clones a face from user photos into generated video and described an opt-out likeness controversy around it. Meta's announcement doesn't describe face-cloning for Muse Video; the episode we can source concerns Muse Image (below). The examples described a product that isn't available to test. This version sticks to what Meta has published.
What Meta has said
From Meta's announcement of Muse Image and Muse Video (2026-07-07):
- Muse Video shares Muse Image's pretraining base and generates audio natively.
- It ranked No. 3 in human-preference Elo for text-to-video on Arena at the time of writing.
- Meta is "investing in areas with current performance gaps, such as audio-video synchronization and physically accurate fast motion."
- It is "coming soon to creators and in Meta AI."
Nothing about resolution, clip length, pricing or an API has been published. Meta's Connect 2026 roundup (2026-09-23) covered its Muse agent, including a Muse Realtime Avatar model for video-chatting with an agent's avatar, which is a different product from the Muse Video generation model.
Muse Image and the Instagram feature Meta pulled
Muse Image launched the same day and works from existing photos too: Meta's examples include placing your pet in a famous painting or combining a selfie with a vacation photo. On July 10, Meta updated its post to say a feature that allowed "@-mentioning public Instagram accounts" for image generation "missed the mark, so it's no longer available". That's the likeness question to keep in mind for any Meta model that works from people's photos.
When you'd use Muse Video
Once it ships, the natural fit is content made and posted inside Meta's apps: short clips with sound for Instagram and Facebook, from Meta AI. For anything you need to produce, price or run through an API today, it isn't an option yet.
What to use meanwhile
On 8frame, video with native audio today:
| Need | Model | Credits |
|---|---|---|
| A shot with sound in one pass | Veo 3.1, 8 s | 448 with audio (Fast 168, Lite 54) |
| Long takes with sound | Seedance 2.5, up to 30 s | 157 for 5 s at 720p |
| Cheap variants of a still you already have | Kling v3 Standard, 5 s (image-to-video only on 8frame: needs a start image) | 86 with audio, 57 silent |
| Cheap variants from a text prompt alone | Kling 2.6 Pro, 5 s | 95 with audio, 47 silent |
| A still of a person speaking an audio track | Hailuo H3 Max Lip Sync, 5 s at 768p | 54 |
For text-only drafts there is also Wan 3.0 at 480p, 24 credits for 5 seconds; it has no audio setting on 8frame. For any of these, a real person's likeness needs that person's documented consent.
FAQ
Is Meta Muse Video available?
Not as of 2026-09-30. Meta's announcement says it is "coming soon to creators and in Meta AI"; no public API, specs or pricing have been published.
Does Muse Video clone your face from photos?
Meta hasn't said so in its announcement. Muse Image works from your photos (for example, combining a selfie with a vacation photo), and Meta removed a feature that let people @-mention public Instagram accounts for image generation.
How good is Muse Video?
Meta reported a No. 3 rank in human-preference Elo for text-to-video on Arena at announcement and named audio-video sync and fast motion as gaps. Until it's public, there's no way to test it on your own briefs.
Related concepts
Muse Video vs Veo 3.1 compares what's known against a model you can run today. Character consistency in AI covers keeping a person stable across shots.
Sources
- Meta AI, Introducing Muse Image and Muse Video (2026-07-07): ai.meta.com/blog/introducing-muse-image-muse-video-msl, checked 2026-09-30
- Meta Newsroom, Introducing Muse Image (2026-07-07, updated 2026-07-10): about.fb.com/news/2026/07/introducing-muse-image-meta-ai, checked 2026-09-30
- Meta, Everything we announced at Connect 2026: meta.com/blog/meta-connect-2026-everything-we-announced, checked 2026-09-30
- 8frame prices: the 8frame tool registry, 2026-09-30; Kling v3 image-to-video only, Kling 2.6 Pro and Wan 3.0 prices (calculateKlingV3Cost, calculateKling26Cost, calculateWan3Cost): read 2026-10-01
Run video with native audio today while Muse is still coming.
Try it without signing up: the 8frame canvas opens on a real board. The canvas is free and unlimited; generation is paid from $19/month.