ByteDance Seedance 2.0

Unified multimodal video generation with native audio and multi-shot narratives

About model

Seedance 2.0 is ByteDance's unified multimodal audio-video generation model, accepting text, image, audio, and video inputs in combination and generating video with synchronized sound in a single pass. Clips run 4 to 15 seconds at native 480p and 720p resolutions, and a single generation can carry professional multi-shot narrative structure with cross-frame consistency, vivid motion, and expressive facial detail. Reference inputs support up to 3 video clips, 9 images, and 3 audio clips per generation, letting the model take motion, style, character, and audio cues directly from source material. ByteDance reports more accurate compliance with complex instructions and more stable motion than earlier Seedance releases. Available on Together AI.

‍

Audio-Video Generation

Native

Dialogue, ambience, and effects synchronized with the visuals

Clip Duration

4-15s

Multi-shot sequences with cross-frame consistency in one generation

Max Reference Inputs

Up to 9 images, 3 video clips, and 3 audio clips per generation

Model key capabilities

Multimodal Reference Inputs: Combine text with up to 9 images, 3 video clips, and 3 audio clips to guide motion, style, characters, and sound
Native Audio-Video Generation: Dialogue, ambient sound, and effects generated in sync with the visuals in a single pass
Multi-Shot Narratives: Professional multi-shot structure with cross-frame consistency, vivid motion detail, and expressive faces
Production-Ready Infrastructure: 99.9% SLA, available on Together AI serverless infrastructure

API usage

cURL
Python
Typescript

Endpoint:

ByteDance/Seedance-2.0

curl --request POST \
  --url https://api.together.xyz/v2/videos \
  --header "Authorization: Bearer $TOGETHER_API_KEY" \
  --header "Content-Type: application/json" \
  --data '{
    "model": "ByteDance/Seedance-2.0",
    "prompt": "some penguins building a snowman"
  }'

from together import Together

client = Together()

# Create a video generation job
job = client.videos.create(
    prompt="A serene sunset over the ocean with gentle waves",
    model="ByteDance/Seedance-2.0"
)

import Together from "together-ai";

const together = new Together();

async function main() {
  // Create a video generation job
  const job = await together.videos.create({
    prompt: "A serene sunset over the ocean with gentle waves",
    model: "ByteDance/Seedance-2.0"
  });

Model card
Architecture Overview:
• Unified multimodal architecture processing text, image, audio, and video inputs for joint audio-video generation
• Native output at 480p and 720p with clip durations from 4 to 15 seconds
• Supports up to 3 video clips, 9 images, and 3 audio clips as reference inputs per generation

Training Methodology:
• Built by the ByteDance Seed team on the group's full-stack generative media research, alongside the Seedream image series and Seed-VL multimodal models

Performance Characteristics:
• ByteDance reports more accurate compliance with complex instructions and more stable motion performance than earlier Seedance releases
• ByteDance reports a competitive usability rate in real-world production settings
‍
Prompting
Together AI API Access:
• Access Seedance 2.0 via Together AI APIs using the endpoint ByteDance/Seedance-2.0
• Authenticate using your Together AI API key in request headers
• Create a video request with a text prompt, optionally combining image, audio, and video reference inputs, then poll until the video completes
• Reference inputs support up to 9 images, 3 video clips, and 3 audio clips per generation
• Available on Together AI serverless infrastructure
‍
Applications & use cases
Film & Narrative Production:
• Generate multi-shot sequences with natural cuts and consistent characters from a single request
• Previsualize scenes with synchronized dialogue, ambience, and effects before full production
• Reference existing footage and audio to carry motion and rhythm into new generations

Marketing & Social Content:
• Produce short-form campaign video with sound in one generation through the Together API
• Match brand style by passing reference images alongside the text prompt
• Batch-generate variations for different platforms and aspect ratios

Product & Brand Video:
• Turn product photography into dynamic clips with matched sound design
• Keep product detail consistent across shots using image references
• Score generations against supplied audio to align pacing with a brand track
‍

Related models

Model specifications

Model data

Model provider
ByteDance
Type
Video
Resolution/Duration
480p-720p, 4-15s
Deployment
Serverless
Endpoint
ByteDance/Seedance-2.0
Price
$0.16 / video estimated per video
Input modalities
Text
Image
Video
Audio
Output modalities
Video

Category
Video

Run in Playground

Quickstart docs

Deploy model

ByteDance Seedance 2.0

About model

API usage

Model card

Prompting

Applications & use cases