> ## Documentation Index
> Fetch the complete documentation index at: https://docs.laozhang.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Gemini Omni Adds 4K at 43% of Google's Price, vs Muse

> Gemini Omni 1.2 Flash now outputs 4K at 43% of Google's list price, and Muse adds built-in styles and camera moves at a flat price. Prices, use cases, requests.

From November 20, 2026, Gemini Omni 1.2 Flash on LaoZhang API generates 4K video at \$0.09 per second, 43% of Google's list price at any length. For built-in styles, camera moves, or first and last frames, use Muse at \$0.30 per 5–10 second clip. Both use `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 `default` group

## Key facts

| Item | Gemini Omni 1.2 Flash | Muse |
| - | - | - |
| Model ID | `gemini-omni-1.2-flash` | `muse-video` |
| Made by | Google | Meta |
| Highest resolution | 4K (3840×2160); 360p, 720p, and 1080p also available | 720p tier (1280×720 at 16:9) |
| Length | 3–10 seconds | 5, 8, or 10 seconds |
| Price | \$0.10, \$0.30, \$0.45 per clip (360p, 720p, 1080p); 4K \$0.09 per second | \$0.30 per clip at any length |
| Audio | Built-in track; characters speak the lines in your prompt | Built-in music and sound effects |
| Image input | Up to 6 images in total, first frame plus reference images | First frame, last frame, up to 4 reference images |
| Aspect ratio | 16:9, 9:16 | 16:9, 9:16, 1:1, 4:3, 3:4 |
| Strengths | Spoken lines, 4K | 7 built-in visual styles, 13 camera moves and transitions |
| Endpoint and key | `POST /v1/videos`, `default` group | Same |

<video controls preload="none" poster="/images/samples/gemini-omni-sample-dew-drops-4k-5s.jpg" src="https://media.laozhang.ai/samples/video/gemini-omni-dew-drops-4k-5s.mp4" />

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

<video controls preload="none" poster="/images/samples/muse-video-sample-anime-rooftop-10s.jpg" src="https://media.laozhang.ai/samples/video/muse-anime-rooftop-10s.mp4" />

muse-video, 10 seconds, 16:9, built-in style 2d\_anime

## What each model is good for

| Use case | Model and setting | Result and price |
| - | - | - |
| 4K footage for large screens | Omni, `"resolution": "4k"` | 3840×2160; \$0.45 for 5 s, \$0.90 for 10 s |
| Talking heads, product pitches, dialogue | Omni, lines written into `prompt` | Characters speak the lines; 10 s at 720p for \$0.30 |
| Testing prompts in bulk | Omni, `"resolution": "360p"` | \$0.10 per clip, usually ready in 30–60 seconds |
| Anime, xianxia, comic, or product-ad looks | Muse, `style_preset` | One field changes the look; \$0.30 for 10 s |
| Short-drama hooks, product-ad shots | Muse, `skill_preset`, for example `micro_drama_hook` or `tvc_product_hero` | Camera moves and transitions cost nothing extra; \$0.30 for 10 s |
| First and last frame control, or 1:1 and 4:3 | Muse, `image` plus `last_frame`, `aspect_ratio` | \$0.30 for 10 s |

## 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):

| 10-second clip | Google list price | LaoZhang API |
| - | -: | -: |
| 360p | \$0.23 | \$0.10 |
| 720p | \$0.70 | \$0.30 |
| 1080p | \$1.04 | \$0.45 |
| 4K | \$2.09 | \$0.90 |

* 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.

For both models, failed tasks are refunded in full, and requests rejected at submission are not charged.

## How to use them

### 4K with Gemini Omni

Change `resolution` to `4k` in your existing request:

```bash theme={null}
curl -s -X POST "https://api.laozhang.ai/v1/videos" \
  -H "Authorization: Bearer $LAOZHANG_API_KEY" \
  -H "Content-Type: application/json" \
  --data-raw '{
    "model": "gemini-omni-1.2-flash",
    "prompt": "Macro shot of dew drops on a spider web at sunrise, shallow depth of field, gentle breeze",
    "resolution": "4k",
    "seconds": "5"
  }'
```

* 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](https://docs.laozhang.ai/en/api-capabilities/gemini-omni-video-generation.md#4k-video).

### A built-in style with Muse

Pass a style ID in `style_preset`, for example 2D anime:

```bash theme={null}
curl -s -X POST "https://api.laozhang.ai/v1/videos" \
  -H "Authorization: Bearer $LAOZHANG_API_KEY" \
  -H "Content-Type: application/json" \
  --data-raw '{
    "model": "muse-video",
    "prompt": "A girl stands on a city rooftop at dusk, wind lifts her hair, she turns and smiles at the camera",
    "style_preset": "2d_anime",
    "seconds": "10",
    "aspect_ratio": "16:9"
  }'
```

All style and camera-move IDs, and first and last frames, are covered under [Muse styles and camera moves](https://docs.laozhang.ai/en/api-capabilities/muse-video-generation.md#set-a-visual-style-and-camera-moves).

## 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 use `1080p` or `4k`, and is not charged. For 2K, use [MiniMax H3](https://docs.laozhang.ai/en/api-capabilities/minimax-h3-video-generation.md).

### 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`.

Other video models are compared in [Video API pricing: H3, Muse, Seedance, Wan](https://docs.laozhang.ai/en/api-capabilities/video-generation-guide.md).

## Sources and related links

* [Gemini Omni 1.2 Flash video generation API](https://docs.laozhang.ai/en/api-capabilities/gemini-omni-video-generation.md)
* [Muse video generation API](https://docs.laozhang.ai/en/api-capabilities/muse-video-generation.md)
* [Video API pricing: H3, Muse, Seedance, Wan](https://docs.laozhang.ai/en/api-capabilities/video-generation-guide.md)
* [Gemini Omni 1.2 Flash launch announcement](https://docs.laozhang.ai/en/announcements/gemini-omni-1-2-flash-2026-11.md)
* [Google: Gemini API pricing](https://ai.google.dev/gemini-api/docs/pricing) (Gemini Omni 1.2 Flash per-second prices, checked November 20, 2026)


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.