doubao-seedance-2-5-260628

doubao-seedance-2-5-260628

ByteDance’s next-generation flagship video-generation large model
2026-08-07
Video Generation
Pricing:
$6 /1M tokensstarting from
Bulk order? Contact your manager for exclusive deals

API Overview

Seedance 2.5 is ByteDance’s next-generation flagship video-generation large model. Seedance 2.5 has achieved a breakthrough leap in dimensions such as instruction following, long-form narrative coherence, lifelike realism, lighting and camera movement, and audio-visual alignment. The model not only boasts native direct output of single 30-second videos and high-fidelity temporal extension capabilities, but also, for the first time, supports highly dense reference input featuring up to 50 multimodal elements (text, images, videos, and audio) combined together. Coupled with millisecond- and second-level timestamps, it enables industrial-grade precise local editing—marking this model as an evolutionary step that elevates AI video from a creative card-drawing tool to standardized industrial productivity.

───────────────────────────────────────────────────────────────────

Core Capabilities

Native 30-Second Video Output and Long-Form Storytelling: It can generate high-definition video clips up to 30 seconds long in a single go and supports seamless multi-round temporal extensions. While maintaining high consistency in character appearance, scene lighting, and cinematic language, it perfectly captures the full arc of the plot and emotional progression.

50+ Ultra-Dense Multimodal References: It allows users to upload up to 30 images, 10 video clips, and 10 audio tracks as reference materials at once (covering character ensembles, scene assets, brand VI, 3D wireframes, and sound tones and melodies), significantly reducing the costs associated with custom commercial content creation, including card-drawing and rework.

Millisecond-Level Timestamps and Precise Local Editing: It supports targeted modifications within specific second ranges of the generated video (e.g., keeping the background and character movements unchanged while replacing product packaging or adjusting lighting and camera effects), enabling industrial-grade iterability—“create once, deliver multiple versions.”

Cinematic Quality and Multilingual Audio-Visual Synchronization: The visuals approach cinematic standards in terms of hair strands, pores, water reflections, and camera movements. It natively supports over ten languages and ensures highly accurate audio-visual alignment between prompts and audio tracks.

API Console

Log in to explore more features! Click to Log In

API Analytics

API Reference (2)

API DescriptionAPI EndpointRequest MethodStabilityParameter Description
Seedance-2.5
POST
Stable
View Details
Seedance-2.5 (Retrieve task results)
GET
Stable
View Details

API Pricing

$
ModelDescription302.AI Price

Seedance-2.5

Input does not contain video

$10 / 1M tokens

Seedance-2.5

Input contains video

$6 / 1M tokens