Seedance 2.0 Fast API
Seedance 2.0 Fast 高速视频生成模型，支持文生、图生和多媒体参考生成。
What is Seedance 2.0 Fast
Seedance 2.0 Fast is the high-speed tier of the Seedance 2.0 video family from ByteDance's Seed team. The Roko catalog positions it as a high-speed video generation model that shares the standard edition's configuration envelope at roughly 60% of its per-tier price, with no 1080p/4K output. Third-party sources describe the Seedance 2.0 family as supporting multimodal references, native AV sync, and multi-shot narrative; Fast itself has no separately published official release date or spec sheet.
Roko connects the model through the asynchronous /v1/videos/tasks API on the video-task and openai-video endpoints, with three routes: seedance-2.0-fast-text-to-video, seedance-2.0-fast-image-to-video, and seedance-2.0-fast-reference-to-video. Input modalities cover text, image, video, and audio with audio generation on by default; clips run 4–15 seconds (default 5), resolution is 480p or 720p only (default 720p), and aspect ratios include adaptive (default) plus 16:9, 9:16, 1:1, 4:3, 3:4, and 21:9.
Pricing follows the Roko production catalog: text-to-video, image-to-video, and image- or audio-only references are billed per output second at $0.0627/s (480p) or $0.1331/s (720p); reference runs that include video or mixed references are billed dual-way at max(input, output) + output seconds, at $0.0374/s (480p) or $0.0792/s (720p).
Model pricing
Pricing follows the current production configuration and the billing dimensions supported by each route.
| Route | Price | Billing rule |
|---|---|---|
| 文生 / 图生 / 图片或音频参考 | 480p $0.0627/sec720p $0.1331/sec | 按输出视频秒数 |
| 视频参考 | 480p $0.0374/sec720p $0.0792/sec | 按 max(输入视频秒数, 输出秒数) + 输出秒数 |
Cost examples
Examples use Roko catalog unit prices: text/image/image- or audio-only references at $0.0627/s (480p) and $0.1331/s (720p); video or mixed references billed dual-way at $0.0374/s (480p) and $0.0792/s (720p). Estimated at submit, settled on actual usage at completion.
5s × $0.0627/s = $0.3135
Cheapest default-duration run — ideal for validating prompts at scale.
5s × $0.1331/s = $0.6655
Default resolution and duration — the sweet spot for social clips and multi-candidate A/B.
15s × $0.1331/s = $1.9965
The longest 720p clip, suited to multi-shot narrative drafts.
(max(5, 10) + 10)s × $0.0792/s = 20s × $0.0792 = $1.584
A 5-second video reference with 10-second output: billed as max(input, output) + output seconds.
Model information
Only provider-documented routes, capabilities, and limits implemented by this gateway are listed.
seedance-2.0-fast-text-to-video
- Limits
- —
- Duration
- 4–15s
seedance-2.0-fast-image-to-video
- Limits
- Images 1–2
- Duration
- 4–15s
seedance-2.0-fast-reference-to-video
- Limits
- Images 0–9 · Videos 0–3 · Audio 0–3
- Duration
- 4–15s
Key capabilities
High-speed tier at about 60% of standard pricing
Fast keeps the standard edition's configuration envelope at roughly 60% of its per-tier price: 720p text-to-video at $0.1331/s on Roko — built for bulk drafts and rapid iteration.
Same reference ceilings as the standard edition
Image-to-video accepts 1–2 images; reference-to-video takes up to 9 images, 3 video clips, and 3 audio tracks, with text, image, video, and audio input modalities.
480p/720p dual resolution
Defaults to 720p with a cheaper 480p tier; both are billed per second ($0.0627/s and $0.1331/s for text-to-video), with 4–15 second durations and a 5-second default.
Seven aspect ratios with adaptive default
adaptive fits the input automatically, while 16:9, 9:16, 1:1, 4:3, 3:4, and 21:9 cover landscape, portrait, and social formats in one place.
Main parameters
Parameters come from structured model configuration; unsupported or unconfigured fields are omitted.
modelstringSubmodel ID
Default:seedance-2.0-fast-text-to-video / seedance-2.0-fast-image-to-video / seedance-2.0-fast-reference-to-videopromptstringSeedance 2.0 Fast 高速视频生成模型，支持文生、图生和多媒体参考生成。
Default:RequireddurationintegerOutput video seconds
Default:4–15squalityenumCurrent billing
Default:480p / 720paspect_ratioenumCapability
Default:adaptive / 16:9 / 9:16 / 1:1 / 4:3 / 3:4 / 21:9image_urlsarrayImage to video: 1–2 · Reference to video: 0–9
Default:Varies by routevideo_urlsarrayReference to video: 0–3
Default:Varies by routeaudio_urlsarrayReference to video: 0–3
Default:Varies by routeCapabilities and usage guidance
Key features
- Text to video · seedance-2.0-fast-text-to-video
- Image to video · seedance-2.0-fast-image-to-video
- Reference to video · seedance-2.0-fast-reference-to-video
- 480p / 720p output
- 4–15s Duration
- Generated audio output
- Audio-capable workflow
Limits
- Image to video: Images 1–2
- Reference to video: Images 0–9 · Videos 0–3 · Audio 0–3
- Capability source: https://evolink.ai/zh/seedance-2-0
Limits and notes
- Resolution is limited to 480p/720p — no 1080p/4K; use seedance-2.0 standard when you need higher output.
- Reference runs that include video or mixed references are billed dual-way at max(input, output) + output seconds, costing more than image- or audio-only references.
- Audio generation is on by default; disable it explicitly in the parameters if you need silent output.
- A successful submit only means the task was accepted — generation can fail or be blocked by moderation, and billing follows actual task settlement.
- Output video links expire; download and re-host them in your own storage promptly.
How it works
- Create a video task: POST to /v1/videos/tasks with model set to seedance-2.0-fast, choosing the text-to-video, image-to-video, or reference-to-video route.
- Configure parameters: duration 4–15s, resolution 480p/720p, and aspect ratio (adaptive by default); pass 1–2 images for image-to-video, or up to 9 images, 3 videos, and 3 audio tracks for reference-to-video.
- Submit the task: use the video-task or openai-video endpoint; a successful submit only means the task was accepted and returns a task ID.
- Poll GET /v1/tasks/{task_id} until it succeeds; you can keep submitting other tasks while generation runs.
- Retrieve the video: download the output URL from the task result and re-host it — the link expires and should not be hot-linked long term.
Use cases
Bulk draft iteration
Validate prompts and storyboards with Fast at low cost, then re-run the winners on seedance-2.0 standard for final renders.
Vertical social clips
9:16 and adaptive ratios plus 4–15 second durations fit short-video feeds, product teasers, and campaign assets.
Low-cost multi-candidate A/B
Generate several candidates per prompt and pick by performance; at roughly 60% of standard pricing, parallel variants stay cheap.
How it compares
The comparison below reflects the current Roko production catalog, covering choices inside the Seedance family and against a cross-family speed tier.
vs seedance-2.0 (standard)
Same configuration envelope, but Fast costs about 60% at the same tier and caps at 720p; choose standard for the 1080p/4K ceiling, Fast for bulk iteration.
vs seedance-2.0-mini (lightweight sibling)
Both are cost-down tiers within the family: Fast prioritizes quality and keeps the full 9+3+3 multimodal references, while Mini prioritizes the lowest cost for very large batches with looser quality needs.
vs kling-3.0-turbo (another speed tier)
Both are speed-first tiers: Fast bills per second across two resolutions (480p/720p), while Kling Turbo bills per second and lets image-to-video inherit the input image's ratio — pick by resolution ceiling and unit price.
FAQ
What is the model ID and which routes does Roko expose?
The model ID is seedance-2.0-fast, served through /v1/videos/tasks on the video-task and openai-video endpoints, with three routes: seedance-2.0-fast-text-to-video, seedance-2.0-fast-image-to-video, and seedance-2.0-fast-reference-to-video.
How is it priced?
Text-to-video, image-to-video, and image- or audio-only references bill per output second: $0.0627/s at 480p and $0.1331/s at 720p. Runs with video or mixed references bill dual-way at max(input, output) + output seconds: $0.0374/s at 480p and $0.0792/s at 720p.
Which durations, resolutions, and ratios are supported?
Durations run 4–15 seconds (default 5); resolution is 480p or 720p only (default 720p); ratios include adaptive (default), 16:9, 9:16, 1:1, 4:3, 3:4, and 21:9.
What are the reference limits?
Image-to-video accepts 1–2 images; reference-to-video accepts up to 9 images, 3 video clips, and 3 audio tracks, with text, image, video, and audio input modalities.
How does it differ from the seedance-2.0 standard edition?
The configuration envelope is identical, but Fast costs about 60% of standard at the same tier and tops out at 720p; the standard edition adds 1080p/4K. Choose standard for resolution, Fast for speed and cost.
Why can't I get the video immediately after submitting?
Video generation is asynchronous: a successful submit only means the task was accepted, and you poll the task until generation finishes. Queues can add latency under heavy concurrency.
Is there a free tier?
Roko bills by actual usage at the unit prices above and has no standing free tier dedicated to this model; any platform-side credits follow the current console policy.
Can I use the output commercially?
Commercial use must comply with Roko's terms of service and the upstream model's licensing scope, and you are responsible for ensuring the content does not infringe third-party rights; the latest agreements on both sides govern.
Sources
Upstream family background comes from third-party public sources; Roko-callable parameters, routes, and prices come from the current production catalog. Verified on 2026-10-13.
- Seedance 2.0 specs (third-party aggregator) Third-party family background: multimodal references, native AV sync, multi-shot narrative
- Seedance 2.0 overview (third-party) Third-party: family background for the Seedance 2.0 line
- Roko API video task docs Async submit, task query, and billing lifecycle on Roko
- Roko API model catalog Current routes, prices, and generation config
Models in this family
Explore other available variants from the same model family.
Seedance 2.0 Fast
Seedance 2.5
View model → ByteDanceSeedance 2.0 Mini
View model → ByteDanceSeedance 2.0 Pro
View model →Other video models
Explore more models available through the unified asynchronous video task API.
