Your first Seedance video: Google sign-in, API key, and three requests

SeedanceLink has two forms of access with different jobs. Google sign-in opens your web console, where you manage your account, balance, and API keys. A video request uses an API key in its Authorization header. A successful browser login does not make a command-line request authenticated, and you do not need to copy an OAuth authorization code into your API client.

This walkthrough follows one small text-to-video task from creation to download. It uses Seedance 2.0, an explicit four-second duration, and 480p resolution. Work through the commands in the same terminal session so the API_BASE, API_KEY, and TASK_ID variables remain available. Replace the key placeholder with a key from your own account.

Prepare your account and key

Choose Google sign-in on the customer site, then open the console. Add sufficient balance for the task when recharge is available. The recharge page displays the payment methods currently enabled for your account; signing in alone does not give a funded API balance. If payment is unavailable, do not assume a disabled payment button represents a completed recharge.

Open API keys and create a key in the Seedance-enabled group. Store it privately rather than in a browser URL, public repository, shared screenshot, or client-side application bundle. For your first test, give the key a recognizable name. Use this same key when checking or downloading the task: ownership is tied to the key that created it.

Submit one explicit request

The smallest example below has a standard retail estimate of USD $0.36 before any applicable group multiplier. Specifying both duration and resolution makes the example predictable: omitted values default to five seconds and 720p, which is a different quote. The model ID is an API identifier, so copy it exactly from the model page rather than replacing it with a display name.

API_KEY='YOUR_API_KEY'
API_BASE='https://seedancelink.io'

curl --fail-with-body --request POST "$API_BASE/api/v3/contents/generations/tasks" \
  --header "Authorization: Bearer $API_KEY" \
  --header "Content-Type: application/json" \
  --data '{
  "model": "seedance-2.0-933",
  "content": [
    {
      "type": "text",
      "text": "A paper boat on a calm lake at sunrise."
    }
  ],
  "duration": 4,
  "resolution": "480p",
  "ratio": "16:9"
}'

The response is a task receipt, not an MP4. Save its id and keep the original request with it. The receipt also includes the task state and billing information. Treat a task with an ID as the existing unit of work even if generation is still queued. Sending the same creation request again would create another task; it is not a status check.

Check, then download

Set TASK_ID to the id in your creation response and query the status approximately every 15 seconds. queued means the task is waiting; running means generation is in progress. Only succeeded is ready for the download step. Keep API_KEY and API_BASE from the creation command, and avoid a fast loop that continuously polls the server.

TASK_ID='REPLACE_WITH_RESPONSE_ID'

curl --fail-with-body "$API_BASE/api/v3/contents/generations/tasks/$TASK_ID" \
  --header "Authorization: Bearer $API_KEY"

After succeeded, call the owned content endpoint using the same API key. The download command follows HTTPS redirects and saves the result as video.mp4. Do not add --location-trusted: a redirect is not a reason to send your private key to another host. A task that has already been paid for can still be read or downloaded with its active owner key when the remaining balance is zero.

curl --fail --location --proto-redir '=https' --max-redirs 3 \
  "$API_BASE/api/v3/contents/generations/tasks/$TASK_ID/content" \
  --header "Authorization: Bearer $API_KEY" \
  --output video.mp4

Questions before your first request