Seedance 2.0 API
Seedance 2.0 is ByteDance's second-generation video model. It generates up to 1080p, 15s video with native audio sync. The breakthrough is the @-reference system: upload video, audio, or image references, and the model extracts camera paths, motion patterns, rhythm, and style.EvoLink will enable Seedance 2.0 access as soon as the official API is live. Join Discord for launch updates, the final model ID once it's live, and day-one pricing/quota notes (or get one launch email).Key takeaway: You can integrate with EvoLink today using the same async workflow (submit → poll → fetch). See the API tab for request format. When Seedance 2.0 becomes available, you'll enable it by switching one model value — no code rewrite.
Seedance 2.0:多模态 AI 视频生成
通过文本、图片、视频和音频引用,一次 API 调用即可生成带同步音频的影视级视频。Seedance 2.0 引入 @引用系统,从源素材提取运动、运镜和风格。相比 1.5 Pro 的核心升级:视频和音频引用输入、每次请求最多 9 张图片 + 3 个视频 + 3 个音频引用、视频转视频编辑模式、15 秒最大时长。

Seedance 2.0 API 能做什么?
Seedance 2.0 引用驱动的视频制作
通过 Seedance 2.0 API,上传参考视频即可通过 @引用系统提取运镜运动、动态节奏和转场风格。从一条主视频生成数十个品牌统一的变体——无需重拍。支持每次请求最多 3 个视频引用,可分别提取运镜、运动和风格。
Seedance 2.0 基于引用的音画同步内容
Seedance 2.0 API 支持提供最多 3 个音频轨作为引用。Seedance 2.0 将剪辑、运动能量和场景转场与节拍和节奏对齐。输出包含原生同步音频——对话支持多语言唇型同步,音效匹配画面动作,背景音乐跟随音频引用的情绪。
Seedance 2.0 多引用分镜转视频
Seedance 2.0 支持组合最多 9 张参考图片和文本提示,跨镜头控制构图、角色和环境。Seedance 2.0 以一致的身份、光照和风格填充关键帧之间的间隙,适合广告制作、产品演示序列和动画分镜。
Seedance 2.0 横向对比 — 一个 EvoLink API Key 访问所有模型
Seedance 2.0 以多模态 @引用输入、视频转视频编辑以及主流视频生成模型中最高的图片引用数量领先。
Seedance 2.0 API 多模态 @引用系统
Seedance 2.0 是唯一在单次请求中支持视频、音频和图片引用的模型。上传最多 9 张图片 + 3 个视频 + 3 个音频来控制运镜、运动、节奏和风格——这些能力在 Sora 2、Kling 3.0 或 Seedance 1.5 Pro 中不可用。
Seedance 2.0 API 视频转视频编辑
Seedance 2.0 API 支持编辑现有视频中的特定片段、角色或动作。Seedance 2.0 支持定向 V2V 修改——Sora 2 或 Kling 3.0 不具备该功能。
Seedance 2.0 API EvoLink 高性价比定价
通过 EvoLink 统一 API 接入 Seedance 2.0 API,享受有竞争力的按秒计费。一个 API Key 即可访问 Seedance 2.0、Sora 2、Kling 3.0、Veo 3.1 等——自动路由到最优服务商。
Pricing & Specs (Quick Comparison)
A transparent benchmark across leading video APIs. Seedance 2.0 pricing will be published on launch.
| Metric | Seedance 2.0 | Sora 2 | Kling 3.0 | Veo 3.1 |
|---|---|---|---|---|
| Price (USD / second, incl. audio) | Coming soon | $0.10 (720p) | $0.112 (1080p) | $0.40 (1080p) |
| Typical cost per video | Coming soon | $1.20 / 12s | $1.12 / 10s | $3.20 / 8s |
| Max duration | 15s | 12s | 15s | 8s |
| Max resolution | TBD | 720p (Std) / 1080p (Pro) | 1080p / 4K (tiered) | 1080p / 4K (tiered) |
| Inputs supported | Text, Image, Video, Audio | Text, Image | Text, Image, Video, Audio | Text, Image |
| Reference controls (Video / Audio / Image) | ✓ / ✓ / ✓ | — / — / ✓ | ✓ / ✓ / ✓ | — / — / ✓ |
Notes: Prices are estimates and may vary by tier/region/provider. "Typical cost per video" examples may use different durations/resolutions as shown. Last updated: 2026-02-21.
如何集成 Seedance 2.0 API
通过 EvoLink 统一 API 将多模态视频生成无缝集成到您的应用,支持 T2V、I2V 和 V2V 模式。
第 1 步 — 获取 API Key
Sign up on EvoLink.ai and generate your secure API key from the dashboard. This key authenticates all your requests to the Seedance 2.0 endpoint.
第 2 步 — 提交生成任务
Send a POST request to /v1/videos/generations with your text prompt, image URLs, or video/audio references. Use the references parameter to pass video, audio, and image inputs for the @-reference system. The API processes this asynchronously and returns a task id for tracking.
第 3 步 — 获取视频结果
使用 `task_id` 轮询状态端点或配置 Webhook。完成后您将收到安全 URL,可下载 MP4+AAC 格式的同步音视频文件。
Seedance 2.0 API 能力
多模态视频制作的技术规格
@引用系统
每次请求上传最多 9 张图片 + 3 个视频 + 3 个音频引用。模型从源素材提取运镜路径、运动模式、节奏和风格。
最高 1080p 高清
输出最高 1080p 分辨率视频(16:9 为 1920x1080),确保所有屏幕上的清晰细节。
4-15 秒时长
支持 4 到 15 秒可变视频长度,多镜头一致性支持更长的叙事序列。
文本、图片、视频和音频输入
支持 T2V(文生视频)、I2V(图生视频)和 V2V(视频转视频)生成模式,多模态输入可自由组合。
原生音频与唇型同步
生成同步的对话、音效和背景音乐。支持多语言唇型同步。
商业授权
商业使用权遵循 BytePlus 条款,通过 EvoLink 平台安全启用。
Seedance 2.0 API 常见问题
Everything you need to know about the product and billing.
全部 Seedance API 模型
EvoLink 提供统一 API 访问完整的 Seedance 模型系列:
所有模型共享同一个 Seedance API 端点,通过一个参数即可切换模型。
API Reference
Select endpoint
Authentication
All APIs require Bearer Token authentication.
Authorization:
Bearer YOUR_API_KEY/v1/videos/generationsCreate Video
Seedance 2.0 (seedance-2.0) model supports multiple generation modes including text-to-video, image-to-video, and video-to-video editing with the @-reference system.
Asynchronous processing mode, use the returned task ID to .
Generated video links are valid for 24 hours, please save them promptly.
Request Parameters
modelstringRequiredDefault: seedance-2.0Video generation model name.
seedance-2.0promptstringRequiredPrompt describing what kind of video to generate. Use @-references to assign roles to uploaded files.
Notes
- Limited to 2000 tokens
- @Image1, @Image2... → images from image_urls (in order)
- @Video1, @Video2... → videos from video_urls (in order)
- @Audio1, @Audio2... → audios from audio_urls (in order)
- Example: '@Image1 as first frame, replicate @Video1 camera movement, @Audio1 for BGM rhythm'
@Image1 as first frame, a cinematic drone shot over a coastal city at golden hourimage_urlsarrayOptionalReference image URL list for image-to-video and multimodal generation.
Notes
- Up to 9 images per request
- Max size: 30MB per image
- Formats: .jpeg, .png, .webp, .bmp, .tiff, .gif
- Mixed input total limit: 12 files across all modalities
- Realistic human face uploads are not supported and will be automatically rejected
- URLs must be directly viewable by the server
http://example.com/image1.jpgvideo_urlsarrayOptionalReference video URL list for video-to-video editing and @-reference system.
Notes
- Up to 3 videos per request, total duration 2–15s
- Max size: 50MB per video
- Formats: .mp4, .mov
- Pixel range: 409,600 (480p) – 927,408 (720p)
- Using video references will increase cost
- Realistic human face uploads are not supported and will be automatically rejected
- URLs must be directly accessible by the server
http://example.com/reference.mp4audio_urlsarrayOptionalReference audio URL list for audio-driven generation via @-reference system.
Notes
- Up to 3 audio tracks per request, total duration ≤ 15s
- Max size: 15MB per audio
- Formats: .mp3, .wav
- Aligns cuts, motion energy, and transitions to beat and rhythm
- URLs must be directly accessible by the server
http://example.com/reference.mp3durationintegerOptionalDefault: 5Specifies the duration of the generated video (in seconds).
| Value | Description |
|---|---|
| 4-15 | Any integer value between 4 and 15 seconds |
Notes
- Billing is based on duration; longer durations result in higher costs
5qualitystringOptionalDefault: 720pVideo resolution.
| Value | Description |
|---|---|
| 480p | Lower resolution, lower pricing |
| 720p | Standard definition (default) |
| 1080p | High definition, higher pricing |
720paspect_ratiostringOptionalDefault: 16:9Video aspect ratio.
| Value | Description |
|---|---|
| 16:9 | Landscape video (default for text-to-video) |
| 9:16 | Portrait video |
| 1:1 | Square video |
| 4:3 | Standard video |
| 3:4 | Portrait standard |
| 21:9 | Ultra-wide video |
| adaptive | Adaptive ratio (default for image-to-video) |
'16:9'generate_audiobooleanOptionalDefault: trueWhether to generate audio. Enabling will increase cost.
| Value | Description |
|---|---|
| true | Video includes synchronized audio (voice, sound effects, background music) |
| false | Video is silent |
Notes
- Place dialogue within double quotes in prompt to optimize audio generation
truecallback_urlstringOptionalHTTPS callback address after task completion.
Notes
- Triggered on completion, failure, or cancellation
- Sent after billing confirmation
- HTTPS only, no internal IPs
- Max length: 2048 chars
- Timeout: 10s, Max 3 retries
https://your-domain.com/webhooks/video-task-completed