COMING SOON

Seedance 2.0

ByteDance's flagship AI video model — #1 on Artificial Analysis Video Arena across all categories. 2K resolution, stereo audio, quad-modal input.

SEEDANCE 1.5 PRO VS 2.0

What's New in 2.0

Every major dimension upgraded — from resolution and duration to a fundamentally new multimodal architecture.

FEATURE1.5 PRO2.0
Max Resolution1080p2K
Max Duration12s15s
AudioBasic syncStereo spatial
Image RefsLimitedUp to 9
Video InputNoneUp to 3 clips
Audio InputNoneUp to 3 files
Motion QualityBaseline2x better
SpeedBaseline30% faster

Seedance 2.0 introduces a unified Dual-Branch Diffusion Transformer that generates video and audio in a single pass — unlike 1.5 Pro's separate processing pipelines. This architectural shift enables stereo spatial audio, physics-aware motion, and seamless multi-shot narratives.

CAPABILITIES

Key Features

Quad-Modal Input

Combine text, images, video clips, and audio files in a single generation. The @ Reference System lets you tag up to 12 files for precise creative control.

Stereo Spatial Audio

Native dual-channel audio with spatial positioning. Material-specific sounds, environmental acoustics, and phoneme-level lip sync in 8+ languages.

Director-Level Camera

Control dolly zooms, rack focuses, tracking shots, POV switches, Hitchcock zooms, and orbit movements through natural language or reference video.

Physics-Aware Motion

Training penalizes impossible motion for realistic weight transfer, momentum conservation, and natural body dynamics — even in complex choreography.

Multi-Shot Narratives

Generate coherent multi-shot sequences with natural cuts and transitions. Use the "lens switch" keyword to signal cuts while preserving character continuity.

Video Editing & Extension

Make targeted modifications to existing clips without full regeneration. Extend videos forward or backward, replace characters, and add or remove elements.

BENCHMARKS

#1 Across All Categories

Seedance 2.0 leads the Artificial Analysis Video Arena — the industry's most comprehensive human-preference leaderboard for AI video models.

Text to Video
#1 Elo 1273

Prompt adherence and visual fidelity from text descriptions

Runner-up: SkyReels V4 (1245)
T2V + Audio
#1 Elo 1213

Video quality with native audio co-generation

Runner-up: Kling 3.0 (1187)
Image to Video
#1 Elo 1352

Animation quality from reference image input

Runner-up: Kling 3.0 (1241)
I2V + Audio
#1 Elo 1168

Image animation with synchronized audio

Runner-up: Kling 3.0 (1142)
90%+ First-attempt
success rate
~60s Average
generation time
30% Faster than
1.5 Pro
SPECIFICATIONS

Technical Specs

DeveloperByteDance (Seed Team)
Release DateFebruary 10, 2026
ArchitectureDual-Branch Diffusion Transformer
Max Resolution2K
Max Duration4-15 seconds
Frame Rate24 fps
Aspect Ratios16:9, 9:16, 1:1, 4:3, 3:4, 21:9
AudioDual-channel stereo, spatial positioning
Lip Sync8+ languages (EN, ZH, JA, KO, ES, FR, DE, PT)
Image InputsUp to 9 per generation
Video InputsUp to 3 clips (max 15s each)
Audio InputsUp to 3 files (max 15s, MP3/WAV)
Total ReferencesUp to 12 files simultaneously
Generation Time~60s standard, ~10min for complex
Success Rate90%+ on first attempt

Frequently Asked Questions

AVAILABLE NOW

Start Creating with Seedance 1.5 Pro

While Seedance 2.0 integration is on its way, generate stunning videos today with our current models.

Try Seedance 1.5 Pro