AIREITER
What Muse Video Ext generates
Muse Video Ext makes one 10-second, 720p clip with native audio per request. Start from a prompt alone, or add up to 9 reference images to keep a person, product, or room consistent.
Text or reference images
Send a prompt, or add 1 to 9 images (11 MiB in total) and name them Picture 1, Picture 2 in the prompt. Reference video, reference audio, and first/last frame control are not supported.
Fixed 10 seconds, 720p, with sound
Length and resolution are fixed. Every clip comes with a generated audio track; there is no mute option to change the price.
Seven aspect ratios
1:1, 16:9, 9:16, 4:3, 3:4, 3:2, or 2:3. Leave it on auto to follow the reference images; text-only requests default to 16:9.
Timestamped shots
Write [Shot 2] At 00:04.000 in the prompt to cut to a new scene. In our test the cuts landed within a fraction of a second of the requested times.
Muse Video Ext test results
These clips are from our own tests on November 4, 2026. Length, pixels, and credits are from the delivered files.
Text to video: snow leopard at golden hour
Request: Text only, 16:9, 10 seconds, audio on.
Delivered file: H.264 MP4, 1280x720, 10.0 seconds, AAC audio with wind and footsteps.
Billing: 5 credits. Generation took about 2 minutes 20 seconds.
Reference images: product try-on
Delivered file: H.264 MP4, 720x1280, 10.0 seconds, AAC audio.
What held up: the chest print, the printed name, and the sleeve badge stayed sharp across a push-in and a slide.
Billing: 5 credits. Generation took about 1 minute 40 seconds.
What to make with Muse Video Ext
Try-on and product close-ups
Reference photos of the product and the person, then one handheld shot with a push-in. Prints, small text, and sleeve badges stayed sharp in our tests.
Short multi-scene ads
Give each scene a timestamp and a reference picture, for example a living room, a try-on, a close-up, and a flat-lay.
Cinematic scenes with sound
Wildlife, landscapes, and street scenes at 16:9, with wind, rain, or footsteps generated to match.
Social formats
Lock 9:16 for vertical feeds or 16:9 for landscape instead of auto when the channel is already decided.
Multi-shot prompts: what to expect
Cuts are accurate, but a person can look different after a cut.
Cut times are reliable
In a five-scene test, the four cuts landed at 2.0, 4.0, 6.0, and 8.0 seconds, matching the prompt.
Products stay consistent
Prints, lettering, and logos stayed the same across every shot.
People may drift after a cut
In the five-scene test, the man looked like a different person from the second shot on. Keep to two cuts or fewer when one person must stay recognisable, and repeat his description in every shot.
Brand names can be refused
Prompts that name a real brand or team can fail content review. Failed tasks are not charged.
How to generate with Muse Video Ext
Use the same steps in the playground or through the API.
Add references (optional)
Upload up to 9 images, 11 MiB in total. Name them Picture 1, Picture 2 in the prompt.
Describe the video
Say who or what moves, then camera, light, and sound. For several scenes, write [Shot N] At 00:0x.000 for each cut.
Choose the ratio
Pick one of seven ratios, or leave auto to follow the references. Length and resolution are fixed.
Submit and poll
POST /api/openapi/submit returns the estimated credits immediately. Poll POST /api/openapi/query with the same out_task_id until completed or failed. A clip takes about a minute, and up to several minutes with many references.
How much a clip costs
One flat price per request.
5 credits per clip
Every request produces one 10-second 720p clip with audio, at 5 credits ($0.05). The price does not change with the number of reference images or the aspect ratio.
Failed tasks are free
If a task fails, including a content-review refusal, no credits are charged.
Call Muse Video Ext through the AIReiter API
One asynchronous contract.
Model ID
Send muse_video_ext.
Submit
POST /api/openapi/submit with model, params (prompt, optional image_url, optional aspect_ratio), and a unique out_task_id.
Query
POST /api/openapi/query with the same out_task_id. Completed responses include credits_used and the output video URL.
Not accepted
Do not send video_length, resolution, seed, negative_prompt, a reference video, or reference audio.
Related AI video models
Wan 3.0
Text, first/last frames, mixed references, and documents, up to 30 seconds.
HappyHorse 1.0
Text, image, multi-reference, and source-video editing on one page.
FLUX 3 Video
Text-to-video, ordered keyframes, and source-video extension with native audio.
Veo 3.1
Cinematic text-to-video and image-to-video.
Muse Video Ext FAQ
Common questions about length, price, references, shots, and the API.
/ 01How long is a Muse Video Ext clip?
Always 10 seconds, at 720p, with a generated audio track.
/ 02How many reference images can I use?
Up to 9, 11 MiB in total. Reference video and reference audio are not supported.
/ 03How much does a clip cost?
5 credits ($0.05) per request, whatever the aspect ratio or number of references. Failed tasks are not charged.
/ 04Can I set a first and last frame?
No. Use reference images and describe the motion in the prompt.
/ 05Can I cut between scenes in one clip?
Yes. Write [Shot N] At 00:0x.000 in the prompt. Cut times are accurate, but a person may look different after a cut, so keep to two cuts or fewer when identity matters.
/ 06Why was my prompt refused?
Prompts that name a real brand, team, or logo can fail content review. Rephrase without the name. Refused tasks are not charged.
/ 07How long does generation take?
About a minute for text only, and up to several minutes with many reference images.
/ 08Which API model ID should I use?
Use muse_video_ext with POST /api/openapi/submit, then poll POST /api/openapi/query with the same out_task_id.