# Kling O3 Text-to-Video

> Generate cinematic AI videos up to 15 seconds with native audio, multi-shot control, and physics-accurate motion via API.


## Overview

- **Endpoint**: `https://api.segmind.com/v1/kling-o3-text2video`
- **Model ID**: `kling-o3-text2video`
- **Category**: Text-to-Video Generation
- **Type**: Synchronous (Direct response)
- **Average Latency**: ~180.8s (30-day average)

- **Average Cost**: $1.96 per run (observed across past runs, not a price — see Pricing)
- **Provider**: EACHLABS



## Pricing

| duration | mode | generate_audio | cost |
| --- | --- | --- | --- |
| 3 | std | false | 0.63 |
| 3 | std | true | 0.84 |
| 3 | pro | false | 0.84 |
| 3 | pro | true | 1.05 |
| 4 | std | false | 0.84 |
| 4 | std | true | 1.12 |
| 4 | pro | false | 1.12 |
| 4 | pro | true | 1.4 |
| 5 | std | false | 1.05 |
| 5 | std | true | 1.4 |
| 5 | pro | false | 1.4 |
| 5 | pro | true | 1.75 |
| 6 | std | false | 1.26 |
| 6 | std | true | 1.68 |
| 6 | pro | false | 1.68 |
| 6 | pro | true | 2.1 |
| 7 | std | false | 1.47 |
| 7 | std | true | 1.96 |
| 7 | pro | false | 1.96 |
| 7 | pro | true | 2.45 |
| 8 | std | false | 1.68 |
| 8 | std | true | 2.24 |
| 8 | pro | false | 2.24 |
| 8 | pro | true | 2.8 |
| 9 | std | false | 1.89 |
| 9 | std | true | 2.52 |
| 9 | pro | false | 2.52 |
| 9 | pro | true | 3.15 |
| 10 | std | false | 2.1 |
| 10 | std | true | 2.8 |
| 10 | pro | false | 2.8 |
| 10 | pro | true | 3.5 |
| 11 | std | false | 2.31 |
| 11 | std | true | 3.08 |
| 11 | pro | false | 3.08 |
| 11 | pro | true | 3.85 |
| 12 | std | false | 2.52 |
| 12 | std | true | 3.36 |
| 12 | pro | false | 3.36 |
| 12 | pro | true | 4.2 |
| 13 | std | false | 2.73 |
| 13 | std | true | 3.64 |
| 13 | pro | false | 3.64 |
| 13 | pro | true | 4.55 |
| 14 | std | false | 2.94 |
| 14 | std | true | 3.92 |
| 14 | pro | false | 3.92 |
| 14 | pro | true | 4.9 |
| 15 | std | false | 3.15 |
| 15 | std | true | 4.2 |
| 15 | pro | false | 4.2 |
| 15 | pro | true | 5.25 |

## API Information

This model uses a **synchronous response pattern**:
1. Make a POST request with your parameters
2. Receive the output directly in the response (binary for images/videos/audio, JSON for text)
3. No polling required - response is immediate


### Input Schema

The API accepts the following input parameters:

- **`mode`** (`string`, _optional_):
  Mode
  Select quality mode. Standard for faster generation, Pro for higher quality.
  - Default: `"pro"`
  - Options: "std" (Standard), "pro" (Pro)

- **`prompt`** (`string`, _required_):
  Prompt
  Describe the video content.

- **`duration`** (`string`, _optional_):
  Duration (seconds)
  Length of the output video in seconds.
  - Default: `"5"`
  - Options: "3" (3), "4" (4), "5" (5), "6" (6), "7" (7), "8" (8), "9" (9), "10" (10), "11" (11), "12" (12), "13" (13), "14" (14), "15" (15)

- **`cfg_scale`** (`number`, _optional_):
  CFG Scale
  Prompt adherence strength.
  - Default: `0.5`
  - Range: 0 to 1

- **`shot_type`** (`string`, _optional_):
  Shot Type
  Camera shot type for the video.
  - Default: `"customize"`

- **`voice_ids`** (`array`, _optional_):
  Voice IDs
  Voice IDs for audio.
  - Item type: string

- **`aspect_ratio`** (`string`, _optional_):
  Aspect Ratio
  Set the aspect ratio for the output video.
  - Default: `"16:9"`
  - Options: "16:9" (16:9), "9:16" (9:16), "1:1" (1:1)

- **`multi_prompt`** (`array`, _optional_):
  Multi Prompt
  Array of prompts with durations for multi-segment video.
  - Item type: object

- **`generate_audio`** (`boolean`, _optional_):
  Generate Audio
  Enable to generate synchronized audio with the video.
  - Default: `false`

- **`negative_prompt`** (`string`, _optional_):
  Negative Prompt
  Specify elements to avoid.
  - Default: `"blur, distort, and low quality"`


**Required Parameters Example**:

```json
{
  "prompt": "A majestic golden eagle soaring over snow-capped mountain peaks at sunrise, cinematic wide angle shot, breathtaking natural scenery, ultra detailed"
}
```


**Full Example**:

```json
{
  "mode": "pro",
  "prompt": "A majestic golden eagle soaring over snow-capped mountain peaks at sunrise, cinematic wide angle shot, breathtaking natural scenery, ultra detailed",
  "duration": "5",
  "cfg_scale": 0.5,
  "shot_type": "customize",
  "voice_ids": [],
  "aspect_ratio": "16:9",
  "multi_prompt": [],
  "generate_audio": false,
  "negative_prompt": "blur, distort, and low quality"
}
```


### Output Schema

The API returns a synchronous response based on the model type:

**For Image/Video/Audio Models**:
- Response contains binary data (image/png, video/mp4, audio/mp3)
- Content-Type header indicates the media type
- Save the response body directly to a file

**For Text Models**:
- Response is JSON with the generated text
- Structure varies by model

**HTTP Response Codes**:
- **200 - OK**: Request successful, output in response body
- **400 - Bad Request**: Invalid parameters
- **401 - Unauthorized**: Invalid or missing API key
- **404 - Not Found**: Model not found
- **406 - Not Acceptable**: Insufficient credits
- **429 - Too Many Requests**: Rate limit exceeded
- **500 - Server Error**: Internal server error



## About

### Kling O3 Text-to-Video — AI Video Generation Model

#### What is Kling O3 Text-to-Video?

Kling O3 (Video 3.0 Omni) is the flagship text-to-video model from Kuaishou Technology, launched in February 2026. It transforms natural language descriptions into high-quality video clips up to 15 seconds long, with support for native synchronized audio generation in a single inference pass. Built on the "Omni One" architecture, Kling O3 combines 3D Spacetime Joint Attention with Chain-of-Thought reasoning — enabling the model to think through scene composition, motion physics, and camera behavior before rendering a single frame. The result is cinematic video that respects real-world physics and maintains consistent characters across shots.

Unlike earlier text-to-video systems that treated generation as a simple mapping task, Kling O3 operates more like a film director: it interprets your prompt for narrative intent, decomposes it into scene elements, plans motion paths and lighting, and executes the full sequence — all in one generation.

#### Key Features

- **Native audio generation**: Dialogue, ambient sound, and music are synthesized simultaneously with the video — no post-processing required. Lip-sync, spatial audio, and breath timing emerge naturally from the same generation pass.
- **Multi-shot storyboarding**: Generate up to 6 distinct camera shots with individual prompts and durations in a single API call, enabling full scene narratives without manual stitching.
- **Physics-accurate motion**: Built-in physics engine models gravity, balance, deformation, collision, and inertia for realistic character and object movement.
- **Flexible aspect ratios**: 16:9 for widescreen, 9:16 for mobile/social, and 1:1 for square content.
- **Standard and Pro quality modes**: Standard delivers faster turnaround for drafts and iteration; Pro produces cinematic-grade output for final delivery.
- **Extended duration**: 3 to 15 second clip lengths — significantly longer than most competing models.

#### Best Use Cases

**Content creation and marketing**: Generate product showcase videos, brand storytelling clips, and social media content at scale without a full production team. The character consistency features are especially useful for series content requiring the same subject across multiple videos.

**Film pre-visualization**: Directors and storyboard artists can quickly prototype shot sequences, camera movements, and scene compositions before committing to production shoots.

**E-commerce**: Demonstrate products in context — lifestyle scenes, environments, and dynamic product shots generated from a brief description.

**Education and training**: Create explainer videos with synchronized narration, visualize abstract concepts, or produce multilingual training scenarios using the built-in multi-language audio support (English, Chinese, Japanese, Korean, Spanish).

**Developers and product teams**: Integrate Kling O3 into content pipelines via the Segmind API endpoint at `https://api.segmind.com/v1/kling-o3-text2video`. Average generation time is approximately 75 seconds per clip.

#### Prompt Tips and Output Quality

Write prompts as scene directions, not image descriptions. A strong Kling O3 prompt specifies the **subject**, **action**, **camera movement**, **lighting**, and **mood** — for example: "Medium shot of a woman walking through a neon-lit Tokyo street at night, camera panning slowly, warm glow reflecting on wet pavement, cinematic and moody."

Use the `cfg_scale` parameter to control how closely the output follows your prompt: values around 0.5 are a good default; push toward 1.0 for strict adherence. Use `negative_prompt` to suppress common artifacts — including `blur, distort, low quality, shaky camera` covers most cases. For Pro mode, generation produces sharper motion and better lighting consistency, especially for longer clips above 8 seconds.

## Usage Guide

### How to Use Kling O3 Text-to-Video

Kling O3 is Kuaishou's most capable text-to-video model, designed for creators and developers who need cinematic, physics-accurate video from natural language prompts. Here's how to get the best results.

#### Writing Effective Prompts

Think like a director, not a photographer. Kling O3 responds best to directorial language that describes action, movement, and atmosphere — not just what a scene looks like.

**Good prompt structure**: `[Shot type] of [subject] [action] in [setting], [camera movement], [lighting], [mood/style].`

Example: "Close-up of a chef plating a dish in a modern kitchen, slow push-in, warm overhead lighting, cinematic and precise."

Avoid vague keywords like "beautiful" or "cool." Describe what the camera should see and how it should move.

#### Choosing Mode and Duration

Use **Standard mode** for fast iteration and testing your prompt. Once you're happy with the composition, switch to **Pro mode** for final output — it produces sharper motion, better lighting consistency, and higher visual fidelity, especially for clips over 5 seconds.

For **duration**, start with 5 seconds to validate your prompt, then extend. Clips over 10 seconds require more detailed prompts to maintain coherence across the full length.

#### Using Multi-Prompt for Scene Sequences

Pass an array to `multi_prompt` to generate multi-shot sequences in a single call. Each segment takes a `prompt` and `duration`:

```json
[
  {"prompt": "Aerial view of a coastline at sunrise, camera drifting forward", "duration": "5"},
  {"prompt": "Medium shot of a surfer paddling out, golden light, slow motion", "duration": "5"},
  {"prompt": "Close-up of a wave crashing, water detail, cinematic", "duration": "5"}
]
```

This is ideal for social media reels, explainer videos, and narrative sequences.

#### Audio Generation

Enable `generate_audio: true` for scenes with speech, environmental sounds, or music. Pair with `voice_ids` from the Kling Create Voice endpoint to use specific cloned or preset voices — reference them in your prompt as `<<<voice_1>>>`.

#### Suppressing Artifacts

Use `negative_prompt` to reduce common quality issues. A reliable default: `blur, distort, low quality, shaky camera, watermark`. For scenes with detailed hands or text on screen, add `distorted hands, garbled text` to further improve output consistency.

#### CFG Scale

Keep `cfg_scale` around 0.5 for most use cases. Increase toward 0.8-1.0 when your prompt is very specific and you want strict adherence. Lower values (0.2-0.4) are useful for experimental or artistic outputs where creative variation is welcome.

---

## FAQ

### What is the maximum video length Kling O3 can generate?

Kling O3 supports clips from 3 to 15 seconds in a single generation.

### Does Kling O3 generate audio automatically?

Audio generation is optional. Enable the `generate_audio` parameter to get synchronized dialogue, ambient sound, and music alongside the video. You can also pass custom voice IDs via the `voice_ids` parameter.

### What is the difference between Standard and Pro mode?

Standard mode is faster and costs less — ideal for drafts and iteration. Pro mode produces higher-quality cinematic output with better lighting, motion, and detail, particularly noticeable in clips over 5 seconds.

### Can I generate multi-shot videos with different scenes?

Yes. Use the `multi_prompt` parameter to pass an array of scene segments, each with its own prompt and duration. This supports up to 6 camera cuts in a single generation.

### What aspect ratios are supported?

16:9 (landscape), 9:16 (portrait/mobile), and 1:1 (square). Select based on your target platform.

### How long does generation take?

Average latency is approximately 75 seconds. Longer durations and Pro mode will take slightly more time.

---

## Usage Examples

### cURL

```bash
curl -X POST "https://api.segmind.com/v1/kling-o3-text2video" \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "mode": "pro",
  "prompt": "A majestic golden eagle soaring over snow-capped mountain peaks at sunrise, cinematic wide angle shot, breathtaking natural scenery, ultra detailed",
  "duration": "5",
  "cfg_scale": 0.5,
  "shot_type": "customize",
  "voice_ids": [],
  "aspect_ratio": "16:9",
  "multi_prompt": [],
  "generate_audio": false,
  "negative_prompt": "blur, distort, and low quality"
}'
```

### Python

```python
import requests
import json

api_key = "YOUR_API_KEY"
url = "https://api.segmind.com/v1/kling-o3-text2video"

data = {
  "mode": "pro",
  "prompt": "A majestic golden eagle soaring over snow-capped mountain peaks at sunrise, cinematic wide angle shot, breathtaking natural scenery, ultra detailed",
  "duration": "5",
  "cfg_scale": 0.5,
  "shot_type": "customize",
  "voice_ids": [],
  "aspect_ratio": "16:9",
  "multi_prompt": [],
  "generate_audio": false,
  "negative_prompt": "blur, distort, and low quality"
}

response = requests.post(
    url,
    json=data,
    headers={
        'x-api-key': api_key,
        'Content-Type': 'application/json'
    }
)

if response.status_code == 200:
    # For image/video/audio models, response.content contains the binary data
    with open('output.png', 'wb') as f:
        f.write(response.content)
    print('Generation complete, saved to output.png')
else:
    print(f"Error: {response.status_code}")
    print(response.text)
```

### JavaScript

```javascript
const apiKey = 'YOUR_API_KEY';
const url = 'https://api.segmind.com/v1/kling-o3-text2video';

const data = {
  "mode": "pro",
  "prompt": "A majestic golden eagle soaring over snow-capped mountain peaks at sunrise, cinematic wide angle shot, breathtaking natural scenery, ultra detailed",
  "duration": "5",
  "cfg_scale": 0.5,
  "shot_type": "customize",
  "voice_ids": [],
  "aspect_ratio": "16:9",
  "multi_prompt": [],
  "generate_audio": false,
  "negative_prompt": "blur, distort, and low quality"
};

const response = await fetch(url, {
  method: 'POST',
  headers: {
    'x-api-key': apiKey,
    'Content-Type': 'application/json',
  },
  body: JSON.stringify(data),
});

if (response.ok) {
  // For image/video/audio models, response contains binary data
  const blob = await response.blob();
  const downloadUrl = URL.createObjectURL(blob);

  // Create download link
  const a = document.createElement('a');
  a.href = downloadUrl;
  a.download = 'output.png';
  a.click();

  console.log('Generation complete');
}
```


## Additional Resources

### Documentation

- [Model Playground](https://www.segmind.com/models/kling-o3-text2video)
- [API Documentation](https://www.segmind.com/models/kling-o3-text2video/api)
- [Pricing Details](https://www.segmind.com/models/kling-o3-text2video/pricing)
- [Platform Documentation](https://docs.segmind.com/)