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

# Omni Video Series

> The all-in-one Gemini Omni video models: three model IDs, two ways to set the length, two billing shapes

Omni is Gemini's all-in-one video line: text-to-video, image-to-video and reference-video editing all run on the same model, served on `POST /v1/tasks`. Switching models means changing `model` and nothing else. The three models differ in **how the length is decided** (a duration you send, or one the model picks), **how reference material is supplied**, and **what they are billed on**.

## Available models

<CardGroup cols={2}>
  <Card title="Omni-Flash-Ext" href="/en/api-reference/video/omni/omni-flash-ext">
    `omni-flash-ext` · fixed 4 / 6 / 8 / 10 second clips, 360p to 4K, first-frame or three-image references; billed at a whole-clip price by resolution × duration, or per second of the reference video
  </Card>

  <Card title="Gemini Omni 1.1 Flash" href="/en/api-reference/video/omni/gemini-omni-flash">
    `gemini-omni-1.1-flash` · model-chosen length of 3–10 s, 360p to 4K, first/last frames and up to 10 reference images; billed by resolution tier × seconds
  </Card>

  <Card title="Gemini Omni Flash Preview" href="/en/api-reference/video/omni/gemini-omni-flash">
    `gemini-omni-flash-preview` · model-chosen length of 3–10 s, 720p only, up to 16 reference images and a reference video of up to 24 s; billed by resolution tier × seconds
  </Card>
</CardGroup>

## Choose a model

| Need                                                 | Recommended model                                      | Why                                                                                              |
| ---------------------------------------------------- | ------------------------------------------------------ | ------------------------------------------------------------------------------------------------ |
| An exact clip length                                 | `omni-flash-ext`                                       | `duration` is a request parameter: `4`, `6`, `8` or `10`                                         |
| A price you can compute before submitting            | `omni-flash-ext`                                       | Without a reference video, resolution and duration are the only billing dimensions               |
| A transition between a first and a last frame        | `gemini-omni-1.1-flash`                                | It accepts `first_frame_image` and `last_frame_image`                                            |
| Many reference images for multiple subjects          | `gemini-omni-flash-preview`                            | Up to 16 images; `gemini-omni-1.1-flash` takes 10 and `omni-flash-ext` takes 3                   |
| Several rounds of edits on the same clip             | `gemini-omni-1.1-flash` or `gemini-omni-flash-preview` | `ref_task_id` continues from the previous result without re-uploading anything                   |
| 4K output                                            | `omni-flash-ext` or `gemini-omni-1.1-flash`            | `gemini-omni-flash-preview` outputs 720p only                                                    |
| Reworking existing footage and paying for its length | `omni-flash-ext`                                       | With `video_urls` it is billed per second of the reference video instead of the whole-clip price |

## What the three models share

These are identical across all three:

| Item            | Value                                                                  |
| --------------- | ---------------------------------------------------------------------- |
| Endpoint        | `POST /v1/tasks` (async task, see [Task System](/en/docs/task-system)) |
| Action          | `generate`                                                             |
| Aspect ratios   | `16:9`, `9:16`, default `16:9`                                         |
| Output count    | One video per request                                                  |
| Reference video | At most one                                                            |
| Not supported   | `seed`                                                                 |

The mode is not selected with `action` but by what you send: `prompt` alone is text-to-video, `image_urls` makes it image-to-video, and `video_urls` reworks an existing clip.

## Unified request

```bash theme={"system"}
curl -X POST https://www.qingbo.dev/v1/tasks \
  -H "Authorization: Bearer $WAVE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemini-omni-1.1-flash",
    "action": "generate",
    "prompt": "sunset over the sea, gulls skimming the water, cinematic light",
    "resolution": "720p",
    "aspect_ratio": "16:9"
  }'
```

A successful submission returns a `task_id`. Retrieve the result with [`GET /v1/tasks/{task_id}`](/en/api-reference/task/status), wait inline with [`Prefer: wait`](/en/api-reference/task/submit#in-request-waiting-prefer-wait), or configure a webhook.

## Related

* [Omni-Flash-Ext](/en/api-reference/video/omni/omni-flash-ext)
* [Gemini Omni Flash](/en/api-reference/video/omni/gemini-omni-flash)
* [Video Generation Overview](/en/api-reference/video/overview)
* [Submit Task](/en/api-reference/task/submit)
* [Query Task Status](/en/api-reference/task/status)
* [Task System](/en/docs/task-system)