POST /v1/videos with the same default-group key.
- Published: November 20, 2026
- Last checked: November 20, 2026
- Status: Active; both models run on the
defaultgroup
Key facts
gemini-omni-1.2-flash, 4K (3840×2160), 5 seconds, 16:9. Prompt: Macro shot of dew drops on a spider web at sunrise, shallow depth of field, gentle breeze
muse-video, 10 seconds, 16:9, built-in style 2d_anime
What each model is good for
Pricing
Gemini Omni 1.2 Flash: 43% of Google’s list price
360p, 720p, and 1080p are billed per clip, the same for 3–10 seconds; 4K is billed per second. Compared with Google’s Gemini API list price, which is per second (checked November 20, 2026):- 4K costs $0.09 per second against Google’s $0.209, so 5, 8, and 10 second clips are all 43% of the list price.
- A first frame or reference images cost nothing extra.
Muse: $0.30 per clip, styles and camera moves included
- 5, 8, and 10 second clips all cost $0.30, which is $0.03 per second at 10 seconds.
- Styles, camera moves, aspect ratios, and images cost nothing extra.
- Meta’s own API has no video generation model; LaoZhang API connects Muse separately and prices it per clip.
How to use them
4K with Gemini Omni
Changeresolution to 4k in your existing request:
- For portrait, add
"aspect_ratio": "9:16"to get 2160×3840. - A 5-second 4K clip usually takes 2–7 minutes; keep polling for up to 30 minutes.
- Polling and downloading are covered under Gemini Omni 4K video.
A built-in style with Muse
Pass a style ID instyle_preset, for example 2D anime:
Who is affected
- Existing Gemini Omni users: your requests and prices stay the same; add
"resolution": "4k"when you want 4K. - Existing Muse users: nothing changes.
- Tokens set to Usage-based only: neither model can be called. Switch the billing mode to Usage first or Per-call.
FAQ
Can Gemini Omni output 2K?
No. Gemini Omni 1.2 Flash has four tiers: 360p, 720p, 1080p, and 4K. A 2K request returns HTTP 400 at submission, telling you to use1080p or 4k, and is not charged. For 2K, use MiniMax H3.
How long does a 4K video take?
A 5-second 4K clip usually takes 2–7 minutes, longer than 1080p. The endpoint is asynchronous: after submitting, poll every 5–10 seconds for up to 30 minutes. Do not resubmit while it is running; a resubmission creates a new task and is charged again.Should I use Gemini Omni or Muse?
- Spoken lines, or 1080p and 4K: use Gemini Omni 1.2 Flash.
- Built-in styles and camera moves, first and last frames, or 1:1 and 4:3 frames: use Muse.
- Both use the same key and endpoint, so you only change
model.
Sources and related links
- Gemini Omni 1.2 Flash video generation API
- Muse video generation API
- Video API pricing: H3, Muse, Seedance, Wan
- Gemini Omni 1.2 Flash launch announcement
- Google: Gemini API pricing (Gemini Omni 1.2 Flash per-second prices, checked November 20, 2026)