Some links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
Seedance 2.5 is ByteDance’s audio-and-video generation model for clips from 4 to 30 seconds. You can send text and reference media through an API, choose output settings such as resolution and audio generation, and retrieve the finished video after an asynchronous task completes. There is no single official price per clip: BytePlus charges by completion-token usage, so the amount depends on the generated output and the applicable rate.
This guide explains the BytePlus workflow, supported controls and constraints, how to estimate costs, and what to check before choosing another API provider.
As an Amazon Associate I earn from qualifying purchases.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Scan for outdated or missing drivers - takes under a minute3Clear out junk files and repair common Windows errorsWhat can Seedance 2.5 do through its API?
Seedance 2.5 jointly generates video and audio. Its documented capabilities include text, image, video, and audio inputs; reference-guided generation; editing; and extending an existing clip. ByteDance describes a 30-second storytelling workflow and says a sequence can be extended twice for longer narratives. Those are capabilities, not a guarantee that every prompt or API request will produce a particular scene, duration, or result.
The API is useful when a product or workflow needs to submit generations programmatically rather than create each clip through a consumer-facing interface. You can submit a task, track its progress, and retrieve its output when it finishes. The request can also set generation options, including whether to generate audio, the aspect ratio, duration, and watermark behavior.
#1 Best Overall
ByteDance also highlights white-model control and green-screen editing for production workflows. These are described as capabilities of the model; the available material does not establish a separate API parameter or a specific compositing workflow for either feature. If a production depends on those controls, confirm how they are exposed in the API version and account you plan to use.
How do I use the official BytePlus API?
BytePlus ModelArk uses an asynchronous task flow: submit a generation request, wait for it to finish, then fetch the video URL from the task result. The documented creation endpoint is https://ark.ap-southeast.bytepluses.com/api/v3/contents/generations/tasks. The API reference names the model dreamina-seedance-2-5-260628.
1. Prepare the request
Authenticate using a Bearer API key, and include the model name and content items in the request. Content items may represent text or supported reference media. Optional settings include generate_audio, ratio, duration, and watermark. Set the options deliberately rather than assuming the defaults match your intended output: 720p is the documented default resolution, and if duration is omitted, the model selects it.
Use the API’s documented content format for the media type and reference data you supply. The available facts establish support for text, image, video, and audio inputs, but do not specify all payload schemas, file-size limits, or authentication setup steps. Consult the current BytePlus API reference for those implementation details before building a production client.
2. Create a generation task
Send a POST request to the task-creation endpoint with the Bearer credential and request body. The response includes a task ID. Save that ID: it is the handle you will use to check the generation rather than expecting the video to be returned in the initial response.
3. Poll or use a webhook
Check task status using GET /contents/generations/tasks/{id}, replacing {id} with the returned task ID. Alternatively, configure a webhook to receive task updates. Continue waiting until the task status is succeeded; a submitted task is not itself proof that the video is ready.
Build the waiting behavior into your application. A task may still be in progress when first checked, so avoid treating an unfinished result as a completed generation. The available documentation does not state a fixed completion time, polling interval, or retry policy; choose those according to the current API guidance and your product’s latency needs.
4. Retrieve the result
When the task succeeds, read content.video_url from its result and use that URL to retrieve the generated file. Your integration should handle the distinction between task completion and downloading the actual video: confirm the task succeeded, then fetch and store or deliver the output using the URL provided.
Which settings and media limits should I plan around?
Seedance 2.5 exposes several output controls, but not all workflows allow them to vary independently. In particular, editing and extension workflows can inherit the selected input video’s ratio and duration behavior.
| Setting or input | Documented behavior | Planning note |
|---|---|---|
| Output resolution | 480p, 720p, or 1080p; 720p is the default. | Choose a resolution explicitly when a workflow or cost estimate depends on it. |
| Duration | 4–30 seconds; if omitted, the model selects the duration. | Specify a value when clip length is important, except where the workflow imposes input-video behavior. |
| Aspect ratio | Can be set for generation; adaptive ratio behavior applies to editing, extension, and first-frame or first-and-last-frame workflows. | Editing and extension lock ratio and duration to the selected input video. |
| Reference images | Dreamina Seedance 2.5 accepts 1–30 reference images. | Use references when you need guidance from still images; the documented range does not establish a guarantee of exact reproduction. |
| Reference videos | Up to 10 reference videos, with a combined duration no longer than 30 seconds. | Budget the total reference-video duration across all supplied clips, not just each clip individually. |
| Audio generation | generate_audio=true generates synchronized audio; false produces silent video. |
Set this according to whether the deliverable needs audio, and include the choice in cost and review planning. |
| Watermark | The watermark parameter controls an AI-generated watermark. |
Check the applicable policy and intended use before choosing the setting. |
| 1080p encoding | BytePlus documents 10-bit color depth and H.265/HEVC encoding for 1080p. | Verify that downstream players and editing tools support the resulting format. |
For editing, extension, and first-frame or first-and-last-frame workflows, the model uses adaptive ratio behavior. Editing keeps output duration approximately aligned with the selected input and may require duration=-1. That is a workflow-specific constraint: do not assume a normal duration choice overrides the input-video behavior. Confirm the exact request requirements for the workflow you are implementing.
Windows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallOutdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchThe stated reference limits apply to Dreamina Seedance 2.5. The API information does not establish that every provider exposes identical limits or accepts the same payload format. If you use a gateway or third-party provider, check its own model listing and request documentation rather than transferring BytePlus limits to it.
How much does Seedance 2.5 cost per video?
BytePlus publishes per-million-token rates, not a fixed price for a five-, ten-, or thirty-second clip. Its pricing explanation estimates video price as the token unit price multiplied by token consumption; actual completion-token usage is returned as usage.completion_tokens. Therefore, a clip’s duration alone is not enough to calculate its exact charge from the published rate.
| BytePlus output and input case | Published rate |
|---|---|
| 480p or 720p, without video input | $10.70 per million tokens |
| 480p or 720p, with video input | $6.40 per million tokens |
| 1080p, without video input | $11.70 per million tokens |
| 1080p, with video input | $7.00 per million tokens |
These are BytePlus’s published 2026 rates. The lower listed rates apply to requests with video input; do not treat them as the standard rate for requests that use only text or images. To calculate a completed request, multiply the applicable rate by its completion-token count divided by 1,000,000. For example, the calculation is rate × completion_tokens ÷ 1,000,000; without the actual token count, quoting a precise per-video cost would be guesswork.
Draft mode, when used, bills the draft and final steps separately. Include both stages when estimating the total for a workflow that uses draft mode, and check the returned usage details for actual consumption. The listed rates do not establish every possible account, regional, tax, or billing condition, so verify the rate shown for your own account before committing to a budget.
How do BytePlus, Vercel AI Gateway, and PiAPI pricing compare?
The providers publish prices in different units, which makes a headline rate comparison misleading. BytePlus and Vercel AI Gateway quote per million tokens, while PiAPI quotes per second. A fair choice requires pricing the same intended workload under each provider’s current rules.
| Provider | Published Seedance 2.5 price | Unit and limitation |
|---|---|---|
| BytePlus ModelArk | $10.70 per million tokens for 480p/720p without video input; $6.40 with video input; $11.70 for 1080p without video input; $7.00 with video input. | Token-based; actual cost depends on completion-token consumption. Draft and final steps are billed separately when draft mode is used. |
| Vercel AI Gateway | $10.70 per million generated tokens. | Token-based listing; the cited rate alone does not provide a per-clip price. |
| PiAPI | $0.15 per second at 480p, $0.35 per second at 720p, and $0.80 per second at 1080p. | Per-second listing; its stated unit differs from token-based prices. |
For PiAPI’s published per-second rates, a simple duration multiplication gives a listed generation charge before any other fees or conditions: a five-second clip is $0.75 at 480p, $1.75 at 720p, or $4.00 at 1080p; a ten-second clip is $1.50, $3.50, or $8.00; a thirty-second clip is $4.50, $10.50, or $24.00. These figures are arithmetic from PiAPI’s 2026 per-second rates, not a claim that its billing has no other terms. The available information does not provide token counts for matching BytePlus or Vercel clips, so a directly equivalent total cannot be computed from their unit rates alone.
Before procurement, compare 5-, 10-, and 30-second workloads at the resolutions you actually need. For token-priced services, use each provider’s current calculator or quote and then validate estimates against usage from completed generations. For a per-second service, confirm whether the billed duration is output length and whether other charges or conditions apply.
What should I verify before choosing an API provider?
Price is only one part of a production decision. The available listings establish some rates and BytePlus workflow details, but they do not establish matching terms across providers for retention, regional availability, activation, or every model capability.
Free tools Windows power users keep installed
One-click scans. No signup required.
- Cost at your workload: price representative clips at 5, 10, and 30 seconds, using the required output resolution and input type. Include draft and final stages if you use draft mode.
- Resolution and duration: verify the provider supports the output sizes and clip lengths your application needs. BytePlus documents 480p, 720p, and 1080p and a 4–30-second range.
- Reference inputs: check image and video limits, combined video-reference duration, accepted media formats, and payload requirements for the provider you will call.
- Audio behavior: confirm whether synchronized audio is supported and whether the API can explicitly produce silent video.
- Task handling: determine whether requests are asynchronous, how to check status, whether webhooks are available, and how result retrieval works.
- Watermarks and output format: establish how watermark settings work and whether the encoding is compatible with your downstream pipeline. BytePlus documents H.265/HEVC and 10-bit color for 1080p.
- Data retention and regional access: get current provider terms for retention, processing region, availability, and account activation. The cited price and capability details do not settle these operational questions.
For an official integration, BytePlus is the route with the documented task endpoint and published technical and pricing details summarized here. A gateway or another API provider may fit a different stack or pricing model, but do not assume that a common model name means identical API behavior, regional access, data handling, or output options.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What are common implementation mistakes?
Expecting the POST response to contain a finished video
The generation process is asynchronous. Keep the task ID, poll the status endpoint or receive webhook updates, and only retrieve the video after success. Treating task creation as completion can lead to empty or premature result handling.
Assuming omitted settings have fixed values
720p is the documented default resolution, but omitted duration is model-selected. If a workflow requires a specific duration, ratio, audio setting, or watermark behavior, pass the relevant option explicitly where the workflow permits it.
Applying ordinary generation assumptions to editing
Editing and extension can lock ratio and duration to the input video; editing duration may need duration=-1 and remains approximately aligned with the selected input. Plan the source clip and intended output together rather than expecting independent duration control.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Budgeting from token rates as if they were per-clip prices
A per-million-token rate does not reveal the cost of a particular video until you know its completion-token usage and applicable input/resolution category. Record usage.completion_tokens from completed calls and use the actual account rate for ongoing estimates.
Comparing unlike provider units
A token rate and a per-second rate cannot be ranked by their displayed numbers. Normalize them against the same duration, resolution, reference-input scenario, and audio requirement, then confirm the provider’s billing definition and other terms.
Frequently Asked Questions
Can I use Seedance 2.5 without generating sound?
Yes. BytePlus documents that setting generate_audio to false produces silent video. If audio is enabled, it is generated in sync with the video.
Does Seedance 2.5 accept multiple reference videos?
Dreamina Seedance 2.5 accepts up to 10 reference videos, provided their combined duration does not exceed 30 seconds. This limit is documented for Dreamina Seedance 2.5 and should not automatically be assumed for third-party integrations.
Recommended Free Tools
Can Seedance 2.5 create a video longer than 30 seconds?
The documented output-duration range is 4–30 seconds per generation. ByteDance describes extending a narrative twice, but the available information does not specify a resulting maximum combined runtime.
Is a Seedance 2.5 API key available immediately after signup?
The cited technical and pricing details do not establish account activation timing or eligibility requirements. Check the current provider account setup and regional availability before planning a launch.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
