Video

Video generation usually takes time. Create a task, save the returned task ID, and use that ID to check the result.

Create a task

POST /v1/video/generations
Authorization: Bearer OMROUTER_API_KEY
Content-Type: application/json

curl --silent --show-error \
  'https://omrouter.com/v1/video/generations' \
  --header "Authorization: Bearer $OMROUTER_API_KEY" \
  --header 'Content-Type: application/json' \
  --data "{
    \"model\": \"$OMROUTER_MODEL_ID\",
    \"prompt\": \"A short establishing shot of a city at sunrise\"
  }"

A successful submission returns a task identifier and initial status:

{
  "task_id": "task-id",
  "status": "submitted"
}

Optional fields such as image input, duration, dimensions, frame rate, and seed depend on the model. Follow the selected video model’s documentation.

Poll a task

GET /v1/video/generations/{task_id}
Authorization: Bearer OMROUTER_API_KEY

curl --silent --show-error \
  "https://omrouter.com/v1/video/generations/$TASK_ID" \
  --header "Authorization: Bearer $OMROUTER_API_KEY"

Terminal responses can contain a result URL or an error:

{
  "task_id": "task-id",
  "status": "succeeded",
  "url": "https://...",
  "format": "mp4"
}

How often to check

• Check every few seconds instead of sending requests continuously.
• Set a maximum wait time.
• Stop checking after a failed result and show the error.
• Do not automatically create another task; that can create a duplicate charge.
• Result URLs can expire, so save the result promptly.

Omrouter also provides OpenAI-style /v1/videos routes. If a client uses that endpoint, follow the client’s task parameter format.