# Kling v3 Video - Core.Today AI API

> Kling Video 3.0: Kuaishou's flagship text/image-to-video model generating cinematic videos up to 15 seconds with multi-shot control, native audio, start/end frame images, and a dedicated 4K mode.

- **Provider**: Kuaishou
- **Model ID**: kwaivgi/kling-v3-video
- **Category**: Video Generation
- **Credits**: 2610 per 5s pro (1080p) video, no audio (default) — billed per second: standard 391/s, pro 521/s, 4k 978/s; with audio standard 587/s, pro 782/s (4k unchanged)
- **Speed**: Slow
- **Quality**: Ultra

## Features
- Cinematic videos up to 15 seconds (3-15s range)
- Three quality tiers: standard (720p), pro (1080p, default), and 4k
- Multi-shot control via multi_prompt (up to 6 sequential shots in one generation)
- Optional native audio generation (off by default)
- Start and end frame images for precise image-to-video control

## Use Cases
- Multi-shot mini-trailers and short narrative films from a single prompt
- Cinematic marketing footage with native synchronized sound
- 4K hero shots for premium brand content
- Image-to-video animation anchored by start and end frames
- Long-form (up to 15s) social clips in vertical or square formats

## API Endpoint

Base URL: https://api.core.today/v1
Create Prediction: POST /predictions
Get Status: GET /predictions/{job_id}

## Authentication
Header: X-API-Key: YOUR_API_KEY

## Input Parameters

### Required
- **prompt**: string - Text prompt for video generation. Max 2500 characters

### Optional
- **mode**: string (default: pro) - 'standard' generates 720p, 'pro' generates 1080p, '4k' generates 4K
  Options: standard, pro, 4k
- **duration**: integer (default: 5) - Video duration in seconds (3-15)
- **start_image**: string - First frame image (.jpg/.jpeg/.png, max 10MB, min 300px, ratio 1:2.5 to 2.5:1)
- **end_image**: string - Last frame image. Requires start_image
- **aspect_ratio**: string (default: 16:9) - Aspect ratio. Ignored when start_image is provided
  Options: 16:9, 9:16, 1:1
- **multi_prompt**: string - JSON array of shot definitions for multi-shot mode, e.g. [{"prompt":"...","duration":3}] (max 6 shots, min 1s per shot, total must equal duration)
- **generate_audio**: boolean (default: false) - Generate native audio for the video (raises the per-second price at standard/pro tiers)
- **negative_prompt**: string - Things you do not want to see in the video. Max 2500 characters

## Examples

### 15-Second Cinematic Ride with Audio
A long-form pro-tier shot with native audio enabled
```json
{
  "model": "kwaivgi/kling-v3-video",
  "input": {
    "mode": "pro",
    "prompt": "First-person POV of a roller coaster plunging into the mouth of an erupting volcano. The track spirals down through rivers of glowing orange lava, sparks and embers flying past the camera. The coaster banks hard around a pillar of molten rock, then launches upward through a vent, bursting out of the volcano's crater into a dazzling sunset sky above the clouds. Wind roaring, riders screaming with excitement, the deep rumble of the volcano.",
    "duration": 15,
    "aspect_ratio": "16:9",
    "generate_audio": true
  }
}
```

### Multi-Shot Sequence (Standard, Silent)
Three coherent shots in one generation at the cheapest tier
```json
{
  "model": "kwaivgi/kling-v3-video",
  "input": {
    "mode": "standard",
    "prompt": "A day in the life of a lighthouse keeper on a remote northern island",
    "multi_prompt": "[{\"prompt\":\"Dawn breaks over a rocky island, the lighthouse silhouetted against pink clouds\",\"duration\":3},{\"prompt\":\"The keeper climbs the spiral staircase, lantern light flickering on stone walls\",\"duration\":3},{\"prompt\":\"Aerial pull-back reveals the lighthouse beam sweeping across a stormy sea at dusk\",\"duration\":3}]",
    "duration": 9,
    "aspect_ratio": "16:9",
    "generate_audio": false
  }
}
```

## Response Format
```json
{
  "job_id": "abc123",
  "status": "pending | processing | completed | failed",
  "result": "URL or data (when completed)"
}
```

## Usage Flow
1. POST /predictions with model and input -> receive job_id
2. GET /predictions/{job_id} -> poll until status is completed or failed
3. Result contains output URL(s)

## Tips
- mode defaults to pro (1080p) at 336/s — switch to standard (252/s) for drafts and iteration, it is 25% cheaper
- Audio raises the per-second price by 50% at standard/pro (252→378, 336→504), but 4k already includes audio pricing at 630/s either way
- 4k mode is the most expensive tier (630/s) — reserve it for final hero renders, not exploration
- Use multi_prompt for coherent multi-shot sequences instead of chaining separate generations; shot durations must sum to duration
- start_image anchors the first frame (aspect_ratio is then ignored), and end_image lets you pin the closing frame too

## Documentation
https://replicate.com/kwaivgi/kling-v3-video