> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.empiriolabs.ai/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.empiriolabs.ai/_mcp/server.

# Kling O3

> Video model in Standard or Pro modes with Text-to-Video, Image-to-Video, Reference-to-Video, editing, native sound, and multi-scene transitions.

![Kling O3](https://media.empiriolabs.ai/model-logos/kling-o3.png)

[Kling AI](/providers/kling) · Video Generation

`POST /v1/videos/generations`

Video model in Standard or Pro modes with Text-to-Video, Image-to-Video, Reference-to-Video, editing, native sound, and multi-scene transitions.

## At a glance

| Field               | Value                         |
| ------------------- | ----------------------------- |
| Model id            | `kling-o3`                    |
| Model release date  | 2026-02-05                    |
| Input modalities    | Text, Image, Video, Audio     |
| Output modalities   | Video                         |
| Context window      | -                             |
| Weight precision    | -                             |
| Features            | audio, editing                |
| Native inference    | No                            |
| New                 | No                            |
| Supported endpoints | `POST /v1/videos/generations` |
| Alternate model ids | `kling/o3`                    |

## Pricing

| Charge                 | Spec       | Rate    |
| ---------------------- | ---------- | ------- |
| Standard T2V/I2V       | per second | \$0.168 |
| Standard T2V/I2V Sound | per second | \$0.224 |
| Standard Video Input   | per second | \$0.252 |
| Pro T2V/I2V            | per second | \$0.224 |
| Pro T2V/I2V Sound      | per second | \$0.280 |
| Pro Video Input        | per second | \$0.336 |
| 4K T2V/I2V/Ref         | per second | \$0.525 |

## Example request

```bash
curl https://api.empiriolabs.ai/v1/videos/generations \
  -H 'Authorization: Bearer $EMPIRIOLABS_API_KEY' \
  -H 'Content-Type: application/json' \
  -d '{"model": "kling-o3", "prompt": "sunrise over the ocean", "duration": 6}'
```

## Parameters

| Parameter             | Type    | Required | Default  | Description                                                                                                                                                                                                                                 |
| --------------------- | ------- | -------- | -------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `prompt`              | string  | yes      | -        | Multi-scene: pipe (\|) or newline-separated prompts, optionally prefixed with duration like '5s: scene text'. Up to 6 scenes.                                                                                                               |
| `model_tier`          | enum    | no       | `"pro"`  | standard: cheapest. pro: balanced quality. 4k: highest fidelity, longest render. · Allowed: `standard`, `pro`, `4k`                                                                                                                         |
| `workflow`            | enum    | no       | `"auto"` | auto: detect from inputs. t2v: text-to-video. i2v: image-to-video. video_edit: edit attached video. reference: use reference_images or reference_videos. · Allowed: `auto`, `t2v`, `i2v`, `video_edit`, `reference`                         |
| `aspect_ratio`        | enum    | no       | `"16:9"` | Kling O3 supports landscape, square, and portrait only. · Allowed: `16:9`, `1:1`, `9:16`                                                                                                                                                    |
| `duration`            | number  | no       | `5`      | Total clip length in seconds. A multi-scene prompt splits it across the scenes unless each scene sets its own length. With a video attached the limit is 10 seconds, and video edit follows the length of the source video. · Range: 3 – 15 |
| `sound`               | boolean | no       | true     | Generate native audio with the video. Not used with a video input; keep_original_sound applies there.                                                                                                                                       |
| `keep_original_sound` | boolean | no       | true     | Video edit, and reference with a video, only. Keep audio from the source video.                                                                                                                                                             |
| `image`               | string  | no       | -        | Reference image URL for i2v.                                                                                                                                                                                                                |
| `image_end`           | string  | no       | -        | Optional last-frame image URL for image-to-video.                                                                                                                                                                                           |
| `video`               | string  | no       | -        | Source video URL for video_edit.                                                                                                                                                                                                            |
| `reference_images`    | string  | no       | -        | Comma-separated image URLs for reference workflow.                                                                                                                                                                                          |
| `reference_videos`    | string  | no       | -        | Comma-separated video URLs for reference workflow.                                                                                                                                                                                          |

## Notes

Video model in Standard or Pro modes with text-to-video, image-to-video, reference-to-video, editing, native sound, and multi-scene transitions.

**Uploaded media preprocessing**

* Video inputs are capped to 10 seconds for video-edit and video-reference workflows.
* Uploaded video inputs are normalized to provider-compatible MP4 when needed.
* Kling O3 4K supports text, image, and image-only reference workflows. Use Standard or Pro for video inputs.

---

*Machine-readable schema:* `GET https://api.empiriolabs.ai/v1/models/kling-o3`.