Generate images and video

EntireFeed exposes stable generation tasks such as video.create and image.edit. The catalog below lists the tasks, models, and required inputs currently available through the public API.

Follow the generation journey

1. Choose a model and task, including model-less processing tasks.
2. Estimate the exact request without creating media or spending balance.
3. Create and monitor the generation when you are ready for paid work.
4. Cancel or retry only after reading the current generation state.

Choose a task, then a model when required

Choose a task for the result you need. Tasks that support models list the available choices and each model's input schema. Tasks that do not use a model omit model entirely rather than sending null.

Use Quickstart for a complete request sequence. Detailed generation lifecycle and operating guides build on this section without changing the task contracts shown here.

Estimate the exact request

POST /v1/estimate accepts the same task, optional model, and input object used by generation creation. Estimation creates no media and spends no balance. Use the returned USD amount before any paid POST /v1/generations call.