Seedance 2.0 API: access, API keys and which endpoint to use

Seedance 2.0 is ByteDance's video-generation model. For a first-party integration, start with BytePlus ModelArk; BytePlus LAS is another first-party route with a different endpoint and service contract. Replicate and other hosted providers offer their own APIs. A model name alone does not tell you whose key, billing or request format to use.

Choose the service first, then the exact variant. Standard, Fast and Mini do not have identical output limits. An API key also does not establish that your account has activated the model or can use every feature.

Choose your API route before copying code

The ByteDance model page explains the model. It does not replace the service documentation needed to authenticate and submit a request. Likewise, a hosted model carrying a ByteDance name is not the same endpoint as BytePlus ModelArk.

ModelArk and LAS use different endpoints

The documented ModelArk task-creation address is:

POST https://ark.ap-southeast.bytepluses.com/api/v3/contents/generations/tasks

The LAS documentation gives this operator address:

POST https://operator.las.ap-southeast-1.bytepluses.com/api/v1/contents/generations/tasks

Keep the service's authentication, request fields and result handling together. Do not combine a LAS URL with a ModelArk example merely because both mention Seedance. Check the region required by your account before using an example address. Sources: ModelArk task API, LAS operator documentation.

Which Seedance 2.0 model ID should you choose?

LAS lists Standard under enhanced video generation and Fast/Mini under basic video generation. These are route-specific documented limits: they should not be transferred to another host without checking that host's schema. LAS also specifies a 4–15-second output range for these three variants. Its parameter notes restrict 1080p in reference-image scenarios. Confirm the task-specific constraint rather than relying only on the model table. LAS capabilities and parameters.

If your application needs 1080p or 4K, a Fast or Mini example is the wrong starting point for this LAS contract. If a provider shows a different resolution list, compare its actual endpoint specification before concluding that the underlying routes are equivalent.

How to get a ModelArk API key

Sign in to BytePlus, select the appropriate region and project, open API Key Management and create a key. ModelArk keys are region-specific and scoped to their project; restrictions can also limit model or endpoint access. Store the key in an environment variable rather than embedding it in application code. The documented authentication header is Authorization: Bearer $ARK_API_KEY. Key management, authentication reference.

Next, check model activation separately. The video API documentation lists balance, AI Savings Plan and available resource-pack routes for activation. Read the current activation rules in your account before committing to a purchase; creating a key is only the authentication step. Model activation prerequisites.

If authentication or access fails, check the key's region, project restrictions and activated model before changing providers. Preserve the actual error message so you can distinguish a permission issue from an unsupported parameter.

What happens after you submit a video task?

ModelArk generation is asynchronous: submission creates a task, and you query its status to obtain the video after completion. Your application needs separate handling for submission, waiting, success and failure. A task identifier is not a completed video. Use the Seedance tutorial and the selected route's retrieval documentation for implementation.

For LAS, the documentation describes a result link valid for 24 hours and task-ID retention of seven days. Plan result retrieval and storage around those limits rather than treating a returned link as permanent hosting. Check equivalent retention rules separately when choosing a hosted provider. LAS output and response requirements.

How to compare API pricing

There is no single price that applies to every Seedance 2.0 API route. ModelArk documents token-based video billing, with the model, output settings and video input affecting the estimate; actual consumption is reported in usage.completion_tokens. Campaign prices have eligibility, dates and caps. Check the current ModelArk pricing before budgeting.

LAS documents a duration-based billing framework with resolution conversion factors. Replicate documents pricing for its own hosted model. Compare the same variant, output duration, resolution and reference-video configuration, then check each service's billable unit. A price copied from one provider is not an official price for every route. LAS billing, Replicate model API.

Keep this access decision separate from a detailed pricing comparison. First establish that the route supports your required task; then estimate cost using its current rules.

Before your first integration test

1. Record the service, region and exact model ID.
2. Confirm activation and key permissions for that service.
3. Check the selected variant's task and output limits.
4. Implement task-status handling and result retrieval.
5. Review the billable unit and any expiring pricing terms.

This guide compares documented interfaces. It does not claim a tested generation speed, successful paid API call or universal account eligibility. For Seedance 2.5, use the separate Seedance 2.5 API guide so that model versions and request limits remain distinct.