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

# Create video generation job

## Veo image-to-video

Start image + prompt for high-quality video. (Doc mirrors upstream fields; includes audio/voice/off-peak notes where listed.)

***

## Quick flow

```
1. Submit job → get task_id
2. Poll status → wait until succeeded for a public URL
```

### Basics

* **Endpoint**: `POST /v1/video/generations`
* **Auth**: `Bearer Token` (`Authorization: Bearer {{YOUR_API_KEY}}`)
* **Content-Type**: `application/json`

***

### Request body

| Field             | Type      | Required | Description                                                                                                                                    |
| :---------------- | :-------- | :------- | :--------------------------------------------------------------------------------------------------------------------------------------------- |
| **model**         | `string`  | **Yes**  | e.g. `veo-2.0-generate-001`, `veo-3.0-fast-generate-001`, `veo-3.0-generate-001`, `veo-3.1-fast-generate-preview`, `veo-3.1-generate-preview`. |
| **prompt**        | `string`  | **Yes**  | Motion description, max 5000 chars.                                                                                                            |
| **aspect\_ratio** | `string`  | **Yes**  | Aspect ratio.                                                                                                                                  |
| **images**        | `array`   | **Yes**  | First frame via `input_reference` (Veo 3.1); when used, `seconds` is fixed to `"8"`.                                                           |
| **seconds**       | `integer` | No       | Veo 3 / 3.1: `"4"`, `"6"`, `"8"`; Veo 2: `"5"`–`"8"` (default `"8"`).                                                                          |
| **size**          | `string`  | No       | `720p` (default), `1080p`, `4k` (4K Veo 3+), or pixel sizes like `1280x720`, `1920x1080`.                                                      |

***

### Image rules

* **Formats**: PNG, JPEG, JPG, WebP
* **Size**: ≤ 50MB file (POST body limit 20MB)
* **Aspect**: between **1:4 and 4:1**

***

### Response

**200 OK (submitted)**

| Field           | Type      | Description    |
| :-------------- | :-------- | :------------- |
| **id**          | `string`  | Instance id.   |
| **task\_id**    | `string`  | Poll id.       |
| **status**      | `string`  | e.g. `queued`. |
| **progress**    | `integer` | Percent.       |
| **created\_at** | `integer` | Created at.    |

```json theme={null}
{
  "id": "task_MQKsdsosLYRreob4Z4NzTFxZKtSoVRWZ",
  "task_id": "task_MQKsdsosLYRreob4Z4NzTFxZKtSoVRWZ",
  "object": "video",
  "model": "veo-2.0-generate-001",
  "status": "queued",
  "progress": 0,
  "created_at": 1775116001
}
```

***

### Request example

```json theme={null}
{
  "model": "Veoq3-turbo",
  "prompt": "Contemporary dance, the people in the picture are performing contemporary dance.",
  "images": ["https://example.com/dance_start_frame.png"],
  "duration": 7
}
```

***

### Notes

1. **audio\_type** split (SFX vs speech) applies on Q2 / Q1 models only, per upstream docs.
2. **Off-peak**: Jobs not finished in 48h may auto-cancel with refund; you can cancel earlier.
3. **is\_rec**: System prompt optimization when unsure (+ credits).
4. **Callback signing**: Verify Veo webhook signatures server-side.