SeeDance 2.5 / 2.0 Video Generator
Generate high-quality video with ByteDance Doubao Seedance 2.5 and 2.0 Standard / Fast / Mini. Supports text-to-video, first frame / first & last frame and multimodal reference-to-video; 2.5 goes up to 30 s and the API accepts up to 30 images + 10 videos + 10 audio references, with optional synchronized audio.
SeeDance 2.5 / 2.0 Video Generator
SeeDance 2.5 supports 4–30 s; Smart mode usually picks ~10 s, roughly doubling the 5 s cost.
Automatically generate synchronized narration, sound effects and background music (on by default).
Additionally return the last frame image of the video in the result.
Add an 'AI-generated' watermark to the output video.
Gacha mode: submit multiple tasks with the same settings
- * Example price. Billing token ≈ duration × W × H × 24 / 1024; pre-charged on submit, reconciled by actual usage on completion.
- When duration is 'Smart' (-1), the model picks the output length and you are billed by the actual seconds.
- Prices match the official rate; discounts come from top-up bonus campaigns. Ample concurrency, no queueing.
- SeeDance 2.5: $12.60/M tokens without video input, $7.56/M tokens with a reference video — actual cost may be below the estimate above.
- SeeDance 2.5 goes up to 30 s; cost scales linearly with seconds.
Prices are estimated defaults; failed tasks are not charged. Final billing is by actual token usage.
No videos generated yet
What is SeeDance 2.5 / 2.0
SeeDance is ByteDance Doubao's video generation model family: 2.5 goes up to 30 s and 1080p, the API takes up to 30 reference images + 10 videos + 10 audio clips, and audio can be the sole reference; 2.0 comes in Standard / Fast / Mini. One endpoint, one content schema — roles distinguish first/last frame, reference images, reference video and reference audio.
Text to Video + Synced Audio
Turn a prompt into a clip (4–30 s on 2.5, 4–15 s on 2.0), optionally with synchronized narration, sound effects and background music.
First Frame
Provide one image as the first frame to control the opening and keep character and scene continuity.
First & Last Frame
Provide both first and last frames to precisely control the start and end; the last frame can relay into the next clip.
Multimodal Reference
Freely combine multiple reference images + reference videos + reference audio to keep subject and style consistent; ingest assets first and reference them by asset:// ID.
Why SeeDance 2.5 / 2.0
ByteDance Doubao-grade quality with native audio generation, priced to match the official rate with ample concurrency.
Models & Capabilities
SeeDance 2.5 plus 2.0 Standard, Fast and Mini cover every video capability through a single endpoint.
doubao-seedance-2-5-260628
Seedance 2.5: 4–30 s, up to 1080p; the API accepts up to 30 reference images + 10 videos + 10 audio clips, audio may be the sole reference; first-frame tasks use the adaptive ratio.
doubao-seedance-2-0-260128
Standard variant, better stability, supports first frame / first & last frame and multimodal reference, up to 1080p.
doubao-seedance-2-0-fast-260128
Fast variant, quicker generation, up to 720p (no 1080p).
doubao-seedance-2-0-mini-260615
Mini variant, faster and about half the price of Standard; 480p / 720p only.
Text to Video
Prompt-only generation (4–30 s on 2.5, 4–15 s on 2.0); duration can be -1 for the model's smart pick.
First / Last Frame / Multimodal
First frame ± last frame, or multiple reference images + reference videos + reference audio, preserving subject and style; this page caps inputs for demo purposes, the API accepts more.
Multiple ratios & resolutions
16:9 / 4:3 / 1:1 / 3:4 / 9:16 / 21:9 / adaptive, 480p–1080p.
SeeDance dedicated key
Use a dedicated APIYI SeeDance2-group token, configured separately in Settings.
FAQ
Questions? Contact hi@apiyi.com for more help.
Get started with SeeDance 2.5 / 2.0
Generate ByteDance Doubao-grade video across four capabilities, up to 30 s on 2.5, with native synchronized audio and direct MP4 output.