# Seedance 2.0 Mini - Core.Today AI API

> Lighter, cheaper variant of ByteDance's Seedance 2.0. Native audio, multimodal reference inputs (images/videos/audio), text-to-video and image-to-video, capped at 720p (no 1080p/4K tier).

- **Provider**: ByteDance
- **Model ID**: bytedance/seedance-2.0-mini
- **Category**: Video Generation
- **Credits**: 1280 per second, tiered by resolution and audio generation
- **Speed**: Fast
- **Quality**: Standard

## Features
- Native synchronized audio generation (dialogue, sound effects, background music)
- Multimodal reference inputs — up to 9 reference images, 3 reference videos, 3 reference audio clips
- Text-to-video and image-to-video (first/last frame) generation
- Intelligent duration mode that auto-picks the best clip length
- Budget-tier pricing at up to 720p resolution

## Use Cases
- Budget-friendly social media clips with built-in dialogue and sound
- Quick concept previews before committing to a higher-tier video model
- Character-consistent short clips using reference images across a series
- Motion-transfer edits driven by reference video clips

## 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. Maximum 4000 characters. BytePlus recommends keeping prompts under 600 English words for best results.

### Optional
- **image**: string - Input image for image-to-video generation (first frame). Cannot be combined with reference images.
- **last_frame_image**: string - Input image for last frame generation. Only works if a first frame image is also provided. Cannot be combined with reference images.
- **reference_images**: array (default: []) - Reference images (up to 9) for character consistency, style guidance, and scene composition. Cannot be used together with first/last frame images. You can reference them in your prompt as [Image1], [Image2], etc.
- **reference_videos**: array (default: []) - Reference videos (up to 3, total duration max 15s) for motion transfer, style reference, and editing. Reference them in your prompt as [Video1], [Video2], etc.
- **reference_audios**: array (default: []) - Reference audio files (up to 3, total duration max 15s) for audio-driven generation and lip-sync. Requires at least one reference image or video. Reference them in your prompt as [Audio1], [Audio2], etc.
- **duration**: integer (default: 5) - Video duration in seconds. Set to -1 for intelligent duration (model picks the best length).
- **resolution**: string (default: 720p) - Video resolution.
  Options: 480p, 720p
- **aspect_ratio**: string (default: 16:9) - Video aspect ratio. Set to 'adaptive' to let the model choose the best ratio based on inputs.
  Options: 16:9, 4:3, 1:1, 3:4, 9:16, 21:9, 9:21, adaptive
- **generate_audio**: boolean (default: true) - Generate synchronized audio with the video, including dialogue (use double quotes in prompt), sound effects, and background music.
- **seed**: integer - Random seed. Set for reproducible generation.

## Examples

### Social clip with native dialogue
Generate a short vertical clip with synchronized dialogue and background music for social media.
```json
{
  "model": "bytedance/seedance-2.0-mini",
  "input": {
    "prompt": "A young woman smiles at the camera and says \"This product is amazing!\" Bright studio lighting, gentle zoom-in.",
    "resolution": "720p",
    "aspect_ratio": "9:16",
    "generate_audio": true,
    "duration": 5
  }
}
```

### Character-consistent scene from reference images
Use reference images to keep a character consistent across a generated scene.
```json
{
  "model": "bytedance/seedance-2.0-mini",
  "input": {
    "prompt": "[Image1] walks through a busy night market, neon signs reflecting in the rain-soaked street.",
    "reference_images": [
      "https://example.com/character-ref.jpg"
    ],
    "resolution": "720p",
    "duration": -1
  }
}
```

## 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
- Set duration to -1 to let Seedance 2.0 Mini automatically pick the best clip length instead of a fixed 5 seconds.
- Use double quotes around spoken lines in the prompt when generate_audio is on — the model treats quoted text as dialogue.
- Reference images/videos/audios cannot be combined with first/last frame images — choose one input mode per generation.
- This tier caps out at 720p; use a higher-tier Seedance/Ray model if you need 1080p or 4K output.

## Documentation
https://replicate.com/bytedance/seedance-2.0-mini