Live54 variants

Seedance 2 API — ByteDance Audio-Video Generation with Director Control

Seedance 2 is ByteDance's unified audio-video model for text-to-video, image-to-video, keyframe transitions, reference-driven scenes, editing, and character workflows.Muapi exposes 54 variants through one asynchronous API, with native audio, phoneme-aware lip-sync in 8+ languages, multi-shot director scripting, and pay-as-you-go pricing from $0.09/sec.

Get a Seedance 2 API Key Try Text to Video Try Omni Reference

Seedance 2.5Next-generation controls and premium output2 variants

T2VSoon

2.5

Seedance 2.5 Text to Video

Flagship Seedance 2 model with 4K support. Cinema-grade quality and premium motion control for production-ready content.

Up to 4K

$0.34/sec

View Details

I2VSoon

2.5

Seedance 2.5 Image to Video

Flagship image-to-video at up to 4K resolution. Maximum photorealism and motion precision in the Seedance family.

Up to 4K

$0.34/sec

View Details

Seedance 2.1Current-generation variants for text and image workflows2 variants

T2VSoon

2.1

Seedance 2.1 Text to Video

Next-generation text-to-video at up to 1080p. Enhanced motion quality and photorealism over Seedance 2.0.

480p–1080p

$0.40/sec

View Details

I2VSoon

2.1

Seedance 2.1 Image to Video

Next-generation image-to-video at up to 1080p. Superior motion fidelity and visual consistency over Seedance 2.0.

480p–1080p

$0.40/sec

View Details

SpicyEligible workflows with relaxed content constraints6 variants

T2VSpicy

Spicy

Seedance 2 Spicy Text to Video

Spicy variant — fast queue and a bolder visual style. Priority compute for text-to-video generation.

720p

$0.30/sec

Try Model

T2VSpicy

Spicy

Seedance 2 Spicy Text to Video Fast

Spicy fast variant — quickest T2V with a bolder visual style and priority queue.

720pFast

$0.21/sec

Try Model

I2VSpicy

Spicy

Seedance 2 Spicy Image to Video

Spicy variant — fast queue and a bolder visual style. Priority image-to-video for high-volume workflows.

720p

$0.30/sec

Try Model

I2VSpicy

Spicy

Seedance 2 Spicy Image to Video Fast

Spicy fast variant — fastest image-to-video available with a bolder visual style.

720pFast

$0.21/sec

Try Model

T2VSpicy

Spicy

Seedance 2 Mini Spicy Text to Video

Mini Spicy variant — the fastest, lowest-cost text-to-video with a bolder visual style.

480p–720pFast

$0.08–$0.15/sec

Try Model

I2VSpicy

Spicy

Seedance 2 Mini Spicy Image to Video

Mini Spicy variant — the fastest, lowest-cost image-to-video with a bolder visual style.

480p–720pFast

$0.08–$0.15/sec

Try Model

VIPPriority queue with premium resolution variants21 variants

T2VVIP

VIP

Seedance 2 VIP Text to Video

VIP variant — fast queue and low censorship. Priority compute for text-to-video generation.

720p

$0.30/sec

Try Model

T2VVIP

VIP

Seedance 2 VIP Text to Video Fast

VIP fast variant — quickest T2V with low censorship and priority queue.

720pFast

$0.21/sec

Try Model

I2VVIP

VIP

Seedance 2 VIP Image to Video

VIP variant — fast queue and low censorship. Priority image-to-video for high-volume workflows.

720p

$0.30/sec

Try Model

I2VVIP

VIP

Seedance 2 VIP Image to Video Fast

VIP fast variant — fastest image-to-video available with low censorship.

720pFast

$0.21/sec

Try Model

FLFVIP

VIP

Seedance 2 VIP First & Last Frame

VIP variant — fast queue and low censorship. Priority first/last-frame interpolation.

720p

$0.30/sec

Try Model

FLFVIP

VIP

Seedance 2 VIP First & Last Frame Fast

VIP fast variant — fastest first/last-frame interpolation with low censorship.

720pFast

$0.21/sec

Try Model

ReferenceVIP

VIP

Seedance 2 VIP Omni Reference

VIP variant — fast queue and low censorship. Priority reference-guided video generation.

720p

$0.30/sec

Try Model

ReferenceVIP

VIP

Seedance 2 VIP Omni Reference Fast

VIP fast variant — quickest reference-guided option with low censorship.

720pFast

$0.21/sec

Try Model

T2VVIP

VIP

Seedance 2 VIP Text to Video 1080p

VIP 1080p variant — full HD text-to-video with priority queue and low censorship.

1080p

$0.675/sec

Try Model

T2VVIP

VIP

Seedance 2 VIP Text to Video 1080p Fast

VIP 1080p fast variant — high-speed full HD text-to-video with priority queue and low censorship.

1080pFast

$0.4725/sec

Try Model

I2VVIP

VIP

Seedance 2 VIP Image to Video 1080p

VIP 1080p variant — animate images to full HD video with priority queue and low censorship.

1080p

$0.675/sec

Try Model

I2VVIP

VIP

Seedance 2 VIP Image to Video 1080p Fast

VIP 1080p fast variant — fastest full HD image animation with priority queue and low censorship.

1080pFast

$0.4725/sec

Try Model

ReferenceVIP

VIP

Seedance 2 VIP Omni Reference 1080p

VIP 1080p variant — full HD reference-guided generation with up to 9 images, 3 videos, and 3 audio clips. Priority queue and low censorship.

1080p

$0.675/sec

Try Model

ReferenceVIP

VIP

Seedance 2 VIP Omni Reference 1080p Fast

VIP 1080p fast variant — fastest full HD reference-guided generation with priority queue and low censorship.

1080pFast

$0.4725/sec

Try Model

FLFVIP

VIP

Seedance 2 VIP First & Last Frame 1080p

VIP 1080p variant — full HD first/last-frame interpolation with priority queue and low censorship.

1080p

$0.675/sec

Try Model

T2VVIP

VIP

Seedance 2 VIP Text to Video 4K

VIP 4K variant — ultra-high-resolution 4K video from a text prompt with priority queue and low censorship.

$1.35/sec

Try Model

I2VVIP

VIP

Seedance 2 VIP Image to Video 4K

VIP 4K variant — animate images to ultra-high-resolution 4K video with priority queue and low censorship.

$1.35/sec

Try Model

FLFVIP

VIP

Seedance 2 VIP First & Last Frame 4K

VIP 4K variant — ultra-high-resolution 4K first/last-frame interpolation with priority queue and low censorship.

$1.35/sec

Try Model

ReferenceVIP

VIP

Seedance 2 VIP Omni Reference 4K

VIP 4K variant — ultra-high-resolution 4K reference-guided generation with up to 9 images, 3 videos, and 3 audio clips. Priority queue and low censorship.

$1.35/sec

Try Model

EditVIP

VIP

Seedance 2 VIP Extend Video

VIP extend variant — fast queue and low censorship. Continues an existing Seedance 2.0 video at 720p while preserving visual style, motion, characters, and audio consistency.

720p

$0.21–$0.30/sec

Try Model

EditVIP

VIP

Seedance 2 VIP Extend Video 1080p

VIP 1080p extend variant — full HD continuation of an existing Seedance 2.0 video with priority queue and low censorship. Preserves style, motion, and audio consistency.

1080p

$0.4725–$0.675/sec

Try Model

GlobalInternational variants for multilingual production10 variants

T2VNew

Global

Seedance 2 Text to Video

Global variant. Generate cinematic videos from text prompts with precise motion and photorealistic quality.

720p

$0.25/sec

Try Model

T2VFast

Global

Seedance 2 Text to Video Fast

Global fast variant. Reduced latency text-to-video, ideal for rapid iteration.

720pFast

$0.15/sec

Try Model

I2VNew

Global

Seedance 2 Image to Video

Global variant. Animate any image into a smooth, realistic video clip with full motion control.

720p

$0.25/sec

Try Model

I2VFast

Global

Seedance 2 Image to Video Fast

Global fast variant. High-throughput image animation without sacrificing visual fidelity.

720pFast

$0.15/sec

Try Model

FLFNew

Global

Seedance 2 First & Last Frame

Global variant. Provide a start and end frame; seamlessly interpolates a fluid video between them.

720p

$0.25/sec

Try Model

FLFFast

Global

Seedance 2 First & Last Frame Fast

Global fast variant. Same smooth first-to-last transitions at significantly lower latency.

720pFast

$0.15/sec

Try Model

ReferenceNew

Global

Seedance 2 Omni Reference

Global variant. Use a reference image to guide character appearance across the generated video. No source video required.

720p

$0.30/sec

Try Model

ReferenceFast

Global

Seedance 2 Omni Reference Fast

Global fast variant. Maintain consistent character identity across scenes at high speed.

720pFast

$0.21/sec

Try Model

EditUtility

Global

Seedance 2.0 Watermark Remover

Remove SD 2.0 watermarks via AI inpainting. Flat $0.025 for clips up to 5s, then $0.005/sec beyond that.

NativeFast

$0.025 / $0.005/sec

Try Model

EditNew

Global

Seedance 2 Watermark Remover Pro

Pro watermark removal for Seedance 2.0 videos. Flat $0.065 for clips up to 5s, then $0.013/sec beyond that.

NativeFast

$0.065 / $0.013/sec

Try Model

MiniLower-cost drafts and fast iteration3 variants

T2VNew

Mini

Seedance 2 Mini Text to Video

Fastest and most affordable Seedance 2 variant. Ideal for rapid iteration, e-commerce assets, and high-volume batch workflows at 480p or 720p.

480p–720pFast

$0.08–$0.15/sec

Try Model

I2VNew

Mini

Seedance 2 Mini Image to Video

Animate images to video at the lowest cost in the Seedance 2 family. 1 image = start frame; up to 9 images = multi-reference via @image1–@image9.

480p–720pFast

$0.08–$0.15/sec

Try Model

ReferenceNew

Mini

Seedance 2 Mini Omni Reference

Reference-guided mini-tier generation. Combine up to 9 images, 3 video clips, and 3 audio files with your prompt for character consistency at minimal cost.

480p–720pFast

$0.08–$0.15/sec + video surcharge

Try Model

ChineseChinese-region variants with audio-video generation8 variants

T2VNew

Chinese

Seedance 2.0 Text to Video 480p

Chinese Seedance 2 text-to-video at 480p resolution. Lower cost with low censorship — ideal for draft generation and rapid iteration.

480p

$0.09–$0.15/sec

Try Model

I2VNew

Chinese

Seedance 2.0 Image to Video 480p

Chinese Seedance 2 image-to-video at 480p resolution. Animate images at lower cost with low censorship.

480p

$0.09–$0.15/sec

Try Model

ReferenceNew

Chinese

Seedance 2.0 Omni Reference 480p

Chinese Seedance 2 omni reference at 480p. Reference-guided generation at lower cost with low censorship.

480p

$0.18/sec

Try Model

T2VNew

Chinese

Seedance v2.0 Text to Video

Chinese Seedance 2 text-to-video with low censorship. Generate cinematic videos from text prompts.

720p

$0.15–$0.25/sec

Try Model

I2VNew

Chinese

Seedance v2.0 Image to Video

Chinese Seedance 2 image-to-video with low censorship. Animate any image into a smooth, realistic video clip.

720p

$0.15–$0.25/sec

Try Model

ReferenceNew

Chinese

Seedance v2.0 Omni Reference

Chinese variant with low censorship. Reference-guided generation using a source video + image for strong character consistency.

720p

$0.30/sec

Try Model

EditNew

Chinese

Seedance v2.0 Extend Video

Seamlessly continue an existing Seedance 2.0 video. Preserves visual style, motion, characters, and audio consistency across the new segment.

720p

$0.15–$0.25/sec

Try Model

EditNew

Chinese

Seedance v2.0 Video Edit

Edit existing videos using text prompts and optional reference images. Billing based on input video duration (max 15s).

720p

$0.25–$0.375/sec

Try Model

Face TrainingCharacter identity training and consistency tools2 variants

Face TrainingNew

Face Training

Seedance 2 Character

Generate a reusable character sheet from reference images. Use the returned character ID in any Seedance 2 Omni Reference prompt.

720p

$0.18 flat

Try Model

Face TrainingNew

Face Training

Seedance 2 Omni Reference Train

Train a custom omni reference model on your subject. Returns a character ID for consistent identity across any scene or prompt.

720p

$0.50 flat

Try Model

What is the Seedance 2 API?

Seedance 2 is ByteDance's unified audio-video generation model. It can create scenes from text, animate images, transition between first and last frames, follow multiple references, and apply edits or extensions to existing footage.

The family combines video generation with native dialogue, ambient sound, foley, and phoneme-aware lip-sync in 8+ languages. Director-mode scripting helps creators plan multi-shot sequences instead of generating isolated clips.

Muapi groups the current 54 variants behind one REST contract, so applications can switch between Chinese, Global, VIP, Mini, Spicy, Face Training, 2.1, and 2.5 options without another provider integration.

Seedance 2 capabilities

Multi-modal reference

Guide a generation with reference images and supported source assets to keep characters, scenes, and style coherent.

Native audio generation

Generate dialogue, ambience, foley, and music in the same audio-video pass.

First and last frame control

Define the opening and closing images for a controlled visual transition.

Director-mode scripting

Describe multi-shot sequences, camera movement, timing, and scene changes in one prompt.

1080p VIP resolution

Use premium variants when you need higher-resolution output and a priority queue.

Seven aspect ratios

Adapt the same generation workflow for landscape, portrait, square, and social-first formats.

Seedance 2 use cases

Multilingual presenters

Create talking-head and character videos with synchronized speech across supported languages.

Product and brand films

Combine product references, camera directions, sound design, and multiple shots in a single workflow.

Storyboards to scenes

Turn scripts and keyframes into coherent clips for pitches, previsualization, and social content.

Character continuity

Use omni reference or face training to keep a recurring person or character recognizable.

Video repair and cleanup

Edit footage, extend a shot, or remove a watermark with a text instruction.

Compare Seedance 2 variants

Variant group	Resolution	Highlights	Best for
Chinese	480p–720p	Core audio-video generation	General regional workflows
Global	720p	Reference and fast international variants	Multilingual production
VIP	720p–4K	Priority queue and low-censorship variants	Premium, high-volume generation
2.1 / 2.5	720p	Newer Seedance family variants	Current-generation experiments

Quick start — Seedance 2 API code examples

Submit a prompt and generation settings, receive a request ID immediately, and poll the standard prediction result endpoint for the completed video URL.

Submit a video task

curl -X POST https://api.muapi.ai/api/v1/seedance-2-text-to-video \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"prompt":"A filmmaker walks through a rain-soaked neon market","resolution":"720p","duration":8,"aspect_ratio":"16:9"}'

# Response: {"request_id":"REQUEST_ID"}

Start with the text-to-video endpoint; image, keyframe, reference, and VIP variants use the same Muapi task lifecycle.

Poll the result

curl https://api.muapi.ai/api/v1/predictions/REQUEST_ID/result \
  -H "x-api-key: YOUR_API_KEY"

# Poll until status is completed, then read the generated video URL.

Poll until the task is completed, then read the generated video URL and any synchronized audio output.

Frequently asked questions

What is the Seedance 2 API?

Seedance 2 is ByteDance's unified audio-video generation model on Muapi. It supports text-to-video, image-to-video, first/last-frame control, omni reference, multi-shot scripting, editing, and synchronized audio.

How many Seedance 2 variants does Muapi offer?

Muapi currently exposes 54 Seedance 2 model variants across Chinese, Global, VIP, Mini, Spicy, Face Training, 2.1, and 2.5 groups.

What is Seedance 2 Omni Reference?

Omni Reference lets you supply multiple reference images so the model can maintain character appearance, scene composition, and style across a generation without a separate fine-tuning workflow.

Does Seedance 2 support lip-sync?

Yes. Seedance 2 includes native audio and phoneme-aware lip-sync for 8+ languages, making it useful for dubbing, localization, and multilingual video.

Can Seedance 2 edit or extend video?

Yes. The catalog includes video edit, extend, and watermark-removal variants that accept a source video and a text instruction.

How do I get Seedance 2 API access?

Create a Muapi account, generate an API key, and call the endpoint matching your workflow. Access is available without a separate waitlist.

Explore Seedance 2.5 →See the next-generation Seedance catalog with 4K and extended multimodal references.Compare Wan 3.0 →Browse another unified video family with text, image, reference, and audio workflows.Create an API key →Start calling Seedance through Muapi.

Ready to direct with Seedance 2?

Use one API key for audio-video generation, references, keyframes, and production variants.

Get API Key Try Seedance 2