BytePlus Seedance 2.5 API: Setup and Pricing 2026

Seedance 2.5 Documentation: BytePlus API Guide 2026

For Seedance 2.5 documentation, begin with BytePlus ModelArk’s official tutorial, then use its create-task API reference for request fields and task-specific constraints. The workflow is to activate the model, create an API key, submit an asynchronous task, poll its status and download the completed video. For browser creation, start with Seedance 2.5; for this website’s separate API, use the Seedance API documentation.

Which documentation answers your question?

Keep provider-specific examples together. A model name, request body and authentication flow should come from the same API contract. A browser model label is not automatically the model identifier required by an API. Likewise, website credits and ModelArk token billing describe separate account systems.

Seedance 2.5 documentation for BytePlus setup

BytePlus ModelArk’s API documentation records the ModelArk identifier dreamina-seedance-2-5-260628. It describes Bearer authentication and the Southeast endpoint https://ark.ap-southeast.bytepluses.com/api/v3. Check the official tutorial when configuring a new integration so the model and endpoint match the account you are using.

1. Create a BytePlus account and open ModelArk.
2. Create an API key and keep it on your server, for example in ARK_API_KEY.
3. Check the current activation conditions and activate Seedance 2.5 in the console.
4. Submit a request to POST /contents/generations/tasks.
5. Save the returned task id and retrieve that task’s status.
6. When the task succeeds, download the output from content.video_url.

BytePlus ModelArk’s documentation records a qualifying resource pack with remaining quota as an activation route. The current create-task documentation also lists qualifying account balance and savings-plan routes; check which condition applies to your account. Verify the current account and purchase pages before deployment. Keep credentials out of browser JavaScript and shared examples. If you need help with the website’s credentials instead, follow the Seedance API key guide.

Request documentation and a minimal example

This example follows the request structure in BytePlus ModelArk’s API documentation. It is an illustrative request, not a report of an executed generation. Set your server environment variable before running it, and verify the model identifier and field compatibility against the current official reference.

curl https://ark.ap-southeast.bytepluses.com/api/v3/contents/generations/tasks \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $ARK_API_KEY" \
  -d '{
    "model": "dreamina-seedance-2-5-260628",
    "content": [{"type": "text", "text": "A ceramic cup rests on a wooden table. Slow camera push-in, soft window light and quiet room ambience."}],
    "generate_audio": true,
    "ratio": "16:9",
    "duration": 5,
    "resolution": "720p",
    "watermark": true
  }'

The create request returns a task rather than the completed video. Use the returned identifier when retrieving status. BytePlus ModelArk’s documentation describes queued and running as in-progress states, and succeeded, failed, cancelled and expired as terminal states. Handle failure separately from a successful result before attempting to read the video URL.

curl "https://ark.ap-southeast.bytepluses.com/api/v3/contents/generations/tasks/$TASK_ID" \
  -H "Authorization: Bearer $ARK_API_KEY"

Save the task identifier, request settings and final status together. This makes a failed request easier to investigate without submitting the same job again merely to discover its state. Download a successful output promptly: BytePlus ModelArk’s documentation describes time-limited output URLs and download restrictions. Do not use a temporary generation URL as your permanent media archive.

Resolution and mode documentation

Seedance 2.5 supports 480p, 720p and 1080p output, with a maximum of 1080p. BytePlus ModelArk’s capability table lists a 4–30-second generation range, audio generation, draft mode, text-to-video, first-frame and first-and-last-frame input, multimodal references, editing and extension. That capability list is a starting point, not permission to combine every field in every task.

The official create-task reference gives special aspect-ratio constraints for Seedance 2.5 editing, extension and first-frame or first-and-last-frame tasks: those cases use adaptive. For editing and extension, it describes preserving the selected input video’s aspect ratio. The text-only example above should therefore not be copied unchanged into a reference or editing workflow.

Check the content type and role for each reference, the intended task type and the output settings as a group. If validation fails, compare the exact field combination with that task’s documentation. Changing the model identifier without reviewing its input constraints is not a complete migration. The related Seedance 2.0 API guide provides version context.

Billing documentation: keep providers separate

The BytePlus pricing guide describes token-based online inference, with unit prices that vary by resolution and whether video is included as input. Its October 19 pricing lists USD 0.0107 per thousand tokens for 480p or 720p without video input, USD 0.0064 with video input, USD 0.0117 for 1080p without video input and USD 0.0070 with video input. These October 19 prices match the official resource-pack rules checked on October 20, 2026. Confirm the final transaction terms before budgeting.

The pricing guide also lists resource-pack examples of 5 million tokens for USD 32, 10 million for USD 64 and 100 million for USD 640, with 90-day validity. Treat these as dated examples, not a checkout quote. An individual request’s cost depends on the applicable billing rules, and a purchased pack has its own validity conditions.

This website’s API instead describes a quote-first workflow: POST /quotes, followed by POST /videos, then GET /generations/{id}. Its Seedance 2.5 identifier is seedance-2.5. See website API models and pricing for that contract, and the video generation API walkthrough for its Python-oriented integration. Do not interchange these routes with ModelArk’s task routes.

Documentation checklist before production

Confirm authentication, activation, the model identifier, input roles, output controls, terminal-state handling and download behavior. Read account quotas and concurrency limits in the provider’s own controls rather than guessing a limit. BytePlus ModelArk’s capability table marks offline flex inference unsupported for Seedance 2.5; do not assume that a feature from another model applies automatically.

For an initial browser evaluation, the site offers free credits on sign-up, enough for one video. Use the Seedance 2.5 trial guide when you need to judge the creative workflow before integrating. A browser trial does not establish API throughput, account quotas or production cost.

FAQ: Seedance 2.5 documentation

Where is the official Seedance 2.5 documentation?

Use the BytePlus ModelArk tutorial and API reference linked above. ByteDance’s model page provides the broader capability overview; the API reference defines request fields and constraints.

Which model ID belongs to ModelArk?

BytePlus ModelArk’s documentation records dreamina-seedance-2-5-260628. Confirm it against the official tutorial for your integration. The website API uses its own identifier.

Can I expect a video in the initial response?

The documented workflow is asynchronous. Save the task identifier, retrieve status and read the video URL after successful completion.

Is 1080p supported?

Seedance 2.5 supports 480p, 720p and 1080p, with a maximum of 1080p. Apply the additional constraints documented for your task type.

Where should I compare API options?

Use the API provider guide and API pricing and limits guide, then verify the selected provider’s current documentation before implementing.