Prices retain their published currencies and billing units below.
Published billing units · original currency · observed 12 Oct 2026
Category
Usage
Rate
Unit
Video generation
480p and 720p output, without reference video1
$5.60
1M video tokens
Video generation
480p and 720p output, with reference video1
$3.30
1M video tokens
Rate conditions
Online generation. Only successful video outputs billed. Actual completion_tokens determines charges; duration, frame rate and resolution affect usage.
Successful videos only. Estimated video tokens = (input-video seconds + output-video seconds) × output width × output height × output fps / 1024. Actual billing uses usage.completion_tokens; these are video accounting units, not language tokens.Source for Billing basis (opens in a new tab)
At most 3 reference clips; each 2–15 seconds; combined at most 15 seconds. Each ≤200 MB, 24–60 fps, 300–6,000 px per side, aspect ratio 0.4–2.5; total pixels 407,696–8,295,044. H.264/H.265; MP4 audio AAC/MP3, MOV also allows PCM.Source for Input video limits (opens in a new tab)
Video-reference requests have a minimum billable video-token floor depending on resolution, aspect ratio and output duration. Exact floor values remain uncollected; use the provider calculator and returned usage.Source for Minimum billable video usage (opens in a new tab)
Pricing coverage
Optional copyrighted-IP use may add a per-IP × output-seconds charge; individual IP rates are shown in the provider experience center and are not collected here.Source for Pricing coverage (opens in a new tab)
Minimum starting balance
Activation guide: available balance ≥CNY200, or a qualifying ≥CNY200 savings plan/resource pack with remaining balance. The API introduction says >CNY200; verify the exact boundary at activation. This is eligibility, not a per-request charge.Source for Minimum starting balance (opens in a new tab)
Prices retain their published currencies and billing units below.
Published billing units · original currency · observed 15 Oct 2026
Category
Usage
Rate
Unit
Native video generation
480p/720p without reference video1
CN¥37.00
1M video accounting tokens
Native video generation
480p/720p with reference video2
CN¥22.00
1M video accounting tokens
Rate conditions
480p/720p output; no reference-video input. Successful output only. Not a fixed per-clip, per-second or language-token price.
480p/720p output; reference-video input included in metered duration and subject to a minimum usage floor. Successful output only. Not a fixed per-clip, per-second or language-token price.
Generation create example omits Bearer while its GET example includes it; use the general authentication convention and confirm with the provider if rejected.
A LAS deployment of the explicitly named Seedance revision with pre/post-processing services. The service enhancement is not a separately identified base-model checkpoint.Source for Release identity (opens in a new tab)
Each video ≤200 MB; 24–60 input fps; sides 300–6,000 px; aspect ratio 0.4–2.5; 409,600–8,295,044 total pixels. Input resolutions include 480p, 720p, 1080p and 4k.Source for Input video limits (opens in a new tab)
Public URLs must remain accessible during execution without login or additional auth headers. Image/audio Base64 is listed; LAS asset IDs require allowlist access.Source for Reference image syntax (opens in a new tab)
Reference asset access
Reference images/videos with real faces must use authorized LAS asset-library IDs after enabling the allowlisted asset feature. Direct face uploads are not supported by the documented operator.Source for Reference asset access (opens in a new tab)
Prompt languages, not a guarantee of generated spoken-language quality.
Prompt limits
Guide recommends no more than 500 Chinese characters or 1,000 English words to reduce ignored details. These are prompt-writing limits, not language-model context tokens.Source for Prompt limits (opens in a new tab)
CNY per combined input/output video second. Estimate = selected rate × (input-video duration + output-video duration); successful outputs only, billed hourly. Actual invoiced usage controls.Source for Billing basis (opens in a new tab)
Pricing coverage
GET still reports video completion_tokens and a minimum-usage floor, while the tariff is per second. No text-token conversion or uncaveated minimum-cost calculator is inferred.Source for Pricing coverage (opens in a new tab)
Streaming
Not collected
Polling and callbacks do not establish streaming generated video.
Batch processing
Not collected
Asynchronous jobs do not establish Batch API support.
Draft preview mode
Not collected
No LAS draft parameter or draft price is established in this contract.
Prices retain their published currencies and billing units below.
Published billing units · original currency · observed 15 Oct 2026
Category
Usage
Rate
Unit
LAS video generation
480p without reference video1
CN¥0.744
combined input/output video second
LAS video generation
480p with reference video2
CN¥0.442
combined input/output video second
LAS video generation
720p without reference video3
CN¥1.598
combined input/output video second
LAS video generation
720p with reference video4
CN¥0.95
combined input/output video second
Rate conditions
480p; no input-video duration. Apply to the sum of input/output video seconds, not output seconds alone. Successful tasks only; actual invoice controls. Separate from Ark video-token tariffs.
480p; input video included in the combined duration. Apply to the sum of input/output video seconds, not output seconds alone. Successful tasks only; actual invoice controls. Separate from Ark video-token tariffs.
720p; no input-video duration. Apply to the sum of input/output video seconds, not output seconds alone. Successful tasks only; actual invoice controls. Separate from Ark video-token tariffs.
720p; input video included in the combined duration. Apply to the sum of input/output video seconds, not output seconds alone. Successful tasks only; actual invoice controls. Separate from Ark video-token tariffs.
Enable LAS and obtain an operator API key. Real-person asset use requires separately enabled LAS Asset Library access. Contract reviewed without invoking the service.
Hosted template workflow on the explicitly selected existing Seedance revision. VLM interpretation, Seedream keyframe editing and structure preprocessing are orchestration stages, not new Seedance base weights.Source for Release identity (opens in a new tab)
Set data.model explicitly to pin this revision. If omitted, the server-configured default applies; the example value is not a guaranteed default.Source for Model selection (opens in a new tab)
basic skips VLM, frame extraction and redraw. general performs keyframe editing. replace/person_replace changes the whole target character. style_transfer/general changes the global visual style while preserving structure.Source for Template behavior (opens in a new tab)
Reference images
basic and style_transfer/general: optional. general: at least one. replace/person_replace: at least one target-person image. No maximum count is stated in this operator schema.Source for Reference images (opens in a new tab)
Reference guidance
Map each source subject to its target reference explicitly in user_prompt. Template paras defaults empty and may contain only fields declared by that template; no complete parameter enum is provided.Source for Reference guidance (opens in a new tab)
Template limitations
Overview mentions face swapping, but the published template table lists only four IDs. No separate face-swap template ID is established. Example style_template=2d_japanese does not establish a full style enum.Source for Template limitations (opens in a new tab)
Fallback behavior
allow_fallback defaults false. If enabled, failed preprocessing may return a basic direct-generation result that does not fulfill the intended semantic edit.Source for Fallback behavior (opens in a new tab)
Video duration
Optional integer duration must be at least −1. At −1 the model selects the output duration. No upper bound is specified by this operator schema.Source for Video duration (opens in a new tab)
Request-table values; the separate 4k tariff is recorded with an availability caveat.
Resolution controls
This editing request table permits 480p and 720p. It does not establish 1080p for this revision; native or generation-only host capabilities are not inherited.Source for Resolution controls (opens in a new tab)
Default adaptive; output reports the actual resolved ratio.
Output formats
Not collected
The operator returns a downloadable video URL. The request table says output_format applies only to 2.5; a sample MP4 URL does not prove a configurable format contract here.
Reference image syntax
Source video accepts public HTTP(S), tos:// paths, asset:// references or a bare asset ID. Image and audio references support public URLs, TOS and LAS assets. TOS buckets must share the primary account and region.Source for Reference image syntax (opens in a new tab)
Reference asset access
LAS Asset Library is allowlisted and disabled by default. The guide requires real-person reference material to use uploaded asset IDs or it may be blocked. data.use_asset_api defaults false; asset_group_id is optional.Source for Reference asset access (opens in a new tab)
Configuration discrepancy
Input prose names seedance_use_asset_api and audio_url; the actual request table and examples use data.use_asset_api and data.audio_urls. Use the request-table fields and confirm account configuration.Source for Configuration discrepancy (opens in a new tab)
Output storage
Overview requires output_tos_path pointing to a writable same-account, same-region TOS directory, but this field is absent from the request table and examples. Placement and requirement remain unresolved.Source for Output storage (opens in a new tab)
Completion callbacks
Submit overview describes callback.url; request table instead exposes data.callback_url as public HTTPS. Callback nesting must be confirmed before relying on notifications.Source for Completion callbacks (opens in a new tab)
task_id is valid for three days. After that it cannot retrieve the processing result. This is job-access lifetime, not an assurance about data deletion or training use.Source for Task record availability (opens in a new tab)
Poll reports the actual model, duration, aspect ratio, resolution and frame rate. template contains only the executed template ID, not the submitted paras. Downstream video states are queued, running, succeeded, failed or expired.Source for Result metadata (opens in a new tab)
Estimated total CNY = 0.16 × output seconds for preprocessing + selected editing rate × (input-video seconds + output-video seconds). Preprocessing depends on template; basic skips preprocessing, but an explicit billing exemption is not stated.Source for Billing basis (opens in a new tab)
Pricing coverage
Successful video outputs only, billed hourly. Invoice controls; source says rounded without specifying rounding precision. usage reports downstream video-token counters, which must not be substituted for the per-second LAS tariff.Source for Pricing coverage (opens in a new tab)
Input limits
Not collected
Source-video duration ceiling, reference byte limits, prompt length and maximum image/audio counts are not specified in this operator contract. Do not inherit unrelated native or LAS-generation limits.
Frame rate
Not collected
Poll reports actual framespersecond; example 24 does not establish a fixed output frame-rate guarantee.
Streaming
Not collected
Polling/callback workflow does not establish streaming video.
Batch processing
Not collected
Concurrent tasks do not establish discounted Batch API support.
Prices retain their published currencies and billing units below.
Published billing units · original currency · observed 15 Oct 2026
Category
Usage
Rate
Unit
Preprocessing add-on
Template-dependent preprocessing1
CN¥0.16
output video second
LAS video editing
480p editing2
CN¥0.442
combined input/output video second
LAS video editing
720p editing3
CN¥0.95
combined input/output video second
Rate conditions
Separate from editing. Applies according to the selected template and output duration. basic skips preprocessing stages, but the guide does not explicitly specify its billing exemption. Rounding granularity is not established; invoice controls.
480p. Multiply by input plus output video duration; add any applicable preprocessing charge. Successful video outputs only. Actual invoice and resolution-dependent usage control; no conversion to chat-token pricing.
720p. Multiply by input plus output video duration; add any applicable preprocessing charge. Successful video outputs only. Actual invoice and resolution-dependent usage control; no conversion to chat-token pricing.
Retention, training use, regional processing and enterprise terms have not been summarized for this endpoint.
Caching
Caching can reuse earlier input processing. Availability, lifetime and billing depend on the endpoint. A cache hit still generates a new response.
01 / INITIAL INPUTProcess & retain
The system processes eligible input and retains reusable state.
02 / LATER REQUESTCheck available state
Reuse depends on matching input, session state and the cache policy.
03 / CACHE HITReuse & generate
Reusing state can reduce repeated work. Any charges follow the billing rules below.
BytePlus ModelArk · standard
dreamina-seedance-2-0-fast-260128
Not verified
How it works
Not collected
Minimum reusable input
Not collected
Lifetime & refresh
Not collected
How to enable it
Not collected
Usage fields
Not collected
Billing rules
Not collected
Limits & data boundaries
Not collected
No verified cache rate for this endpoint. An empty rate is not a zero charge.
Volcengine Ark video · default online inference
doubao-seedance-2-0-fast-260128
Not verified
No native prompt-cache contract or discounted cached-input tariff established. Draft reuse, stored assets and output retention are not prompt caching.
How it works
Not collected
No native prompt-cache contract or discounted cached-input tariff established. Draft reuse, stored assets and output retention are not prompt caching.
Minimum reusable input
Not collected
No native prompt-cache contract or discounted cached-input tariff established. Draft reuse, stored assets and output retention are not prompt caching.
Lifetime & refresh
Not collected
No native prompt-cache contract or discounted cached-input tariff established. Draft reuse, stored assets and output retention are not prompt caching.
How to enable it
Not collected
No native prompt-cache contract or discounted cached-input tariff established. Draft reuse, stored assets and output retention are not prompt caching.
Usage fields
Not collected
No native prompt-cache contract or discounted cached-input tariff established. Draft reuse, stored assets and output retention are not prompt caching.
Billing rules
Not collected
No native prompt-cache contract or discounted cached-input tariff established. Draft reuse, stored assets and output retention are not prompt caching.
Limits & data boundaries
Not collected
No native prompt-cache contract or discounted cached-input tariff established. Draft reuse, stored assets and output retention are not prompt caching.
No verified cache rate for this endpoint. An empty rate is not a zero charge.
Caching reviewed 15 Oct 2026
Volcengine LAS video (basic) · LAS basic video generation
doubao-seedance-2-0-fast-260128
Not verified
No LAS video prompt-cache contract or cached-input discount established. Asset reuse and stored tasks are not prompt caching.
How it works
Not collected
No LAS video prompt-cache contract or cached-input discount established. Asset reuse and stored tasks are not prompt caching.
Minimum reusable input
Not collected
No LAS video prompt-cache contract or cached-input discount established. Asset reuse and stored tasks are not prompt caching.
Lifetime & refresh
Not collected
No LAS video prompt-cache contract or cached-input discount established. Asset reuse and stored tasks are not prompt caching.
How to enable it
Not collected
No LAS video prompt-cache contract or cached-input discount established. Asset reuse and stored tasks are not prompt caching.
Usage fields
Not collected
No LAS video prompt-cache contract or cached-input discount established. Asset reuse and stored tasks are not prompt caching.
Billing rules
Not collected
No LAS video prompt-cache contract or cached-input discount established. Asset reuse and stored tasks are not prompt caching.
Limits & data boundaries
Not collected
No LAS video prompt-cache contract or cached-input discount established. Asset reuse and stored tasks are not prompt caching.
No verified cache rate for this endpoint. An empty rate is not a zero charge.
Caching reviewed 15 Oct 2026
Volcengine LAS editing 2.0
doubao-seedance-2-0-fast-260128
Not verified
No inference-cache mechanism or cache discount documented. Stored assets, tasks and reused references are not prompt caching.
How it works
Not collected
No inference-cache mechanism or cache discount documented. Stored assets, tasks and reused references are not prompt caching.
Minimum reusable input
Not collected
No inference-cache mechanism or cache discount documented. Stored assets, tasks and reused references are not prompt caching.
Lifetime & refresh
Not collected
No inference-cache mechanism or cache discount documented. Stored assets, tasks and reused references are not prompt caching.
How to enable it
Not collected
No inference-cache mechanism or cache discount documented. Stored assets, tasks and reused references are not prompt caching.
Usage fields
Not collected
No inference-cache mechanism or cache discount documented. Stored assets, tasks and reused references are not prompt caching.
Billing rules
Not collected
No inference-cache mechanism or cache discount documented. Stored assets, tasks and reused references are not prompt caching.
Limits & data boundaries
Not collected
No inference-cache mechanism or cache discount documented. Stored assets, tasks and reused references are not prompt caching.
No verified cache rate for this endpoint. An empty rate is not a zero charge.
Caching reviewed 15 Oct 2026
What the measurements say
No benchmark observations have been added for this model. We do not turn a vendor description into a performance score.
Latency and throughput depend on the host, request size, reasoning settings and load. There is no measured speed or global usage figure in this record.
The record behind the model
Why it is included
Inferred from the named native hosted API offering and cross-check of ByteDance and ByteDance-Seed public repositories. No matching downloadable checkpoint was found. This is not an explicit published weight-license declaration.