· reAPI Team
Meta Muse Video Release Date, API and Pricing
When Meta previewed Muse Video, whether there is a Meta Muse Video API yet, what it costs per clip, and what one 10-second 720p clip with sound contains.
Meta previewed Muse Video on July 7, 2026, alongside the launch of Muse Image, and said it was "coming soon to creators and Meta AI"[1]. That date is the one most people mean when they search for the Meta Muse Video release date. It was a preview, not a general release: Meta published sample clips and a leaderboard result, and opened Muse Image to the public, but not Muse Video.
Developers want to know what has happened since. As of October 29, 2026, Meta Muse Video is not on Meta's own Model API[2], but it can be called through reAPI as muse-video[3]. This guide covers the timeline, what Meta has said about the model's strengths and gaps, how the API works, what a clip costs and whether any of it is free.
TL;DR
- Release date: Meta previewed Muse Video on July 7, 2026, and called it "coming soon to creators and Meta AI"[1].
- Meta's own API: Muse Video is not in the Meta Model API model list as of October 29, 2026[2].
- API access today: reAPI serves
muse-videoon/api/v1/videos/generations, about 10 seconds at 720p with sound per clip[3]. - Pricing on reAPI (checked October 29, 2026): $0.024 per clip, 24 credits, not multiplied by seconds[4].
- Quality, by Meta's account: No. 3 on Arena for text to video as of July 5, 2026, with audio-video sync and fast motion named as gaps[1].
- Free? No free API tier; reAPI's 100 signup credits cover four Meta Muse Video clips.
The Meta Muse Video release date and timeline
Meta announced two models on July 7, 2026, the first media generation models from Meta Superintelligence Labs. The research post opens: "We're excited to launch Muse Image and preview Muse Video"[1]. The wording is deliberate. Muse Image went live that day in the Meta AI app, on meta.ai and inside Instagram and WhatsApp features[5]. Muse Video got a section headed "Previewing Muse Video" and the line "Muse Video is coming soon to creators and in Meta AI"[1].
| Date | Event | Source |
|---|---|---|
| July 5, 2026 | Arena standings Meta quotes for Muse Video (No. 3, text to video) | Meta research post[1] |
| July 7, 2026 | Meta previews Muse Video and launches Muse Image | Meta research post[1] |
| October 29, 2026 | Muse Video still absent from Meta Model API's model list | Meta Models page[2] |
| October 29, 2026 | muse-video callable on reAPI | reAPI docs[3] |
Meta has not published a general-availability date for Muse Video in Meta AI. The research post gives no date beyond "coming soon"[1], so any specific day you see quoted for a public launch inside Meta's apps needs its own source.
What Meta says Muse Video does well, and where it falls short
Meta describes Muse Video as "built on the same pretraining base" as Muse Image and says it "delivers exceptional visual fidelity with native audio support"[1]. In the preview section Meta adds that it "offers competitive performance in prompt adherence, visual fidelity, and temporal consistency"[1].
The same paragraph is unusually direct about the gaps: "We're investing in areas with current performance gaps, such as audio-video synchronization and physically accurate fast motion"[1]. On the leaderboard side, Meta reports that "on Arena, Muse Video ranks No. 3 in human-preference Elo for text-to-video at the time of writing", with rankings dated July 5, 2026[1].
Read those two statements together. A clip with dialogue that must land on exact lip movements, or a sports shot with fast physical action, is where Meta itself expects Meta Muse Video to struggle. Ambient scenes, product shots and slow camera moves with sound effects sit closer to what the preview clips show.
The Meta Muse Video API: what you can call today
Meta's developer documentation lists the models on its Model API as Muse Wonder, Muse Spark, Muse Image, Muse Voice Transcribe and Segment Anything, plus two open-weight models[2]. Muse Video is not among them as of October 29, 2026. There is no Meta Muse Video API from Meta to sign up for yet.
reAPI serves the model as muse-video on POST /api/v1/videos/generations[3]. The request surface is small:
| Field | What it takes |
|---|---|
prompt | Required, also with references |
image_urls | 0–9 public image URLs, 11 MB in total |
aspect_ratio | 1:1, 16:9, 9:16, 4:3, 3:4, 3:2 or 2:3; leave it out for auto |
duration | Only 10 |
resolution | Only 720p |
audio | Only true |
Source: reAPI Muse Video docs[3].
Without references, auto gives a 16:9 clip. With references, the clip follows their shape and stays close to them, but its first frame is not an exact copy of an image[3]. There are no first or last frame, video reference, audio reference, extend or seed fields. The call returns a task id; you poll /api/v1/tasks/{id} until the MP4 is ready.
A minimal request looks like this:
curl https://reapi.ai/api/v1/videos/generations \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "muse-video",
"prompt": "A golden retriever shakes off water on a sunny dock, droplets flying in slow motion, the camera pushing in gently, splashes and a happy bark. No text.",
"aspect_ratio": "16:9"
}'Writing prompts for Meta Muse Video
Meta's preview section shows the prompts behind its sample clips, and they share a pattern worth copying. Each one describes a single continuous moment, names the light and camera behaviour, and then gives the sound its own clause. One reads: "A man juggles three oranges, adds a fourth, drops them all, and takes a bow anyway. Warm natural morning light, gentle slow-motion. A single continuous ~10-second moment with a clear beginning, turn, and payoff. AUDIO: quiet room tone with crisp foley"[1].
Three habits follow from those examples:
- Plan for one beat. A clip is about 10 seconds, so write one action with a start, a turn and an end rather than a sequence of scenes.
- Name the sound separately. Ambience, foley and music each get words of their own, because the audio is generated with the picture.
- Stay inside the stated strengths. Slow camera moves and steady subjects suit the model better than fast physical action, the gap Meta itself names[1].
Meta Muse Video pricing
Meta has published no price for Muse Video, because it is not on Meta's API[2]. For comparison only, Meta prices its sibling Muse Image at a flat $0.01 per generated image[6].
On reAPI, Muse Video costs one flat price per clip: $0.024, billed as 24 credits at $1 = 1,000 credits, checked October 29, 2026[4]. The price is not multiplied by seconds, and references or aspect ratio do not change it[3].
| Clips | Cost on reAPI at $0.024 each | Seconds of video (about 10 s each) |
|---|---|---|
| 1 | $0.024 | about 10 |
| 10 | $0.24 | about 100 |
| 100 | $2.40 | about 1,000 |
| 1,000 | $24.00 | about 10,000 |
Arithmetic: clips × $0.024. Dividing $0.024 by a 10-second clip gives about $0.0024 per second of output, a useful number when you compare Meta Muse Video with models that bill by the second.
Failed tasks are refunded in full[3], so the table assumes every clip succeeds and you keep it. If you generate three takes to keep one, multiply by three.
Is Meta Muse Video free?
Not as an API. Meta has said Muse Video is coming to Meta AI, where its sibling Muse Image is "free for everyday creation"[5], but Meta has not said how Muse Video will be offered there. From code, every clip is paid. reAPI gives new accounts 100 credits, worth $0.10, enough for four clips at 24 credits each[4]. That is a test allowance, not a free tier.
Meta Muse Video FAQ
When is the Meta Muse Video release date?
Meta previewed Muse Video on July 7, 2026, and said it was "coming soon to creators and Meta AI"[1]. Meta has not announced a public launch date inside its apps.
Is there a Meta Muse Video API?
Not from Meta yet: Muse Video is not in Meta Model API's model list as of October 29, 2026[2]. You can call it through reAPI as muse-video[3].
How much does Meta Muse Video cost?
On reAPI, $0.024 per clip of about 10 seconds, billed as 24 credits[4]. Meta has published no Muse Video price.
Is Meta Muse Video free?
There is no free API tier. reAPI's 100 signup credits cover four clips[4].
How long is a Muse Video clip?
About 10 seconds. The duration field accepts only 10, and each request returns one clip[3].
Does Muse Video generate sound?
Yes. Meta describes it as having "native audio support"[1], and every clip on reAPI includes audio; audio: false is refused[3].
Is Meta Muse Video any good?
Meta reports No. 3 on Arena for text to video as of July 5, 2026, and names audio-video synchronization and fast physical motion as current gaps[1]. Test it on your own prompts before relying on it.
Where Meta Muse Video stands today
Meta previewed the model on July 7, 2026 and has not yet put it on its own API or given a launch date inside Meta AI. You do not have to wait to test it: on reAPI, Meta Muse Video is callable today as muse-video, one clip of about 10 seconds at 720p with sound per request, at $0.024 a clip. Start with a few prompts in the Muse Video playground, and pair it with Muse Image when you want to animate a still.
References
- AI at Meta. Introducing Muse Image and Muse Video. July 7, 2026. Retrieved October 2026 from ai.meta.com/blog/introducing-muse-image-muse-video-msl
- Meta. Models. Retrieved October 29, 2026 from dev.meta.ai/docs/models
- reAPI. Muse Video API reference. reapi.ai/docs/muse-video
- reAPI. Muse Video model page and pricing. Checked October 29, 2026 at reapi.ai/models/muse-video
- Meta Newsroom. Introducing Muse Image: Image Generation Built for Your World. July 7, 2026. Retrieved October 2026 from about.fb.com/news/2026/07/introducing-muse-image-meta-ai
- Meta. Pricing and rate limits. Retrieved October 2026 from dev.meta.ai/docs/pricing-rate-limits