Videos

Use GET /api/open/v1/media/models?mediaType=video to discover verified models, POST /api/open/v1/media/videos/jobs to create a job, and GET /api/open/v1/media/videos/jobs/{jobId} to poll it.

Request

Create requests include a prompt, model, duration, size, and optional frame or reference media. All requests use bearer authentication:

Authorization: Bearer KITTA_API_KEY
Idempotency-Key: UNIQUE_REQUEST_ID

curl https://kittaai.com/api/open/v1/media/videos/jobs \
  -H "Authorization: Bearer $KITTA_API_KEY" \
  -H "Idempotency-Key: video-$(uuidgen)" \
  -H "Content-Type: application/json" \
  -d '{"prompt":"SUV driving beside a lake","model":"veo3.1","seconds":8,"size":"720x1280"}'

Response

Video task responses usually return a task id and status first, then a result video URL after completion. Store the task id and original project id together, because video generation can outlive a browser session.

Billing And Credits

Video jobs precharge API quota, not website member credits. Provider creation, persistence, async failure, and timeout paths refund the API quota idempotently. Check Profile before creating batches.

Idempotency-Key is required and must be unique for each intended creation. Retrying with the same key returns the original response without creating or charging a second task.

Errors

Handle invalid prompt, unsupported model, unreachable reference media, insufficient credits, and processing failure. If the schema does not expose video endpoints, show the capability as unavailable instead of attempting hidden routes.