Skip to navigation
Ishushka API (actual)

Video generation

View as Markdown

Generate videos asynchronously with the Ishushka API. Use the preparation endpoint to discover and validate the settings supported by a model before starting a billable generation.

Workflow

  1. List video models with GET /v1/models?type=video. The catalog includes request-dependent pricing options; choose the SKU that matches the generation settings.
  2. Call POST /v1/videos/prepare with the model and selected settings. It returns supported values and an estimated customer price. Select resolution first, then aspect ratio, then size. Repeat until is_complete is true.
  3. Call POST /v1/videos/generate with the same settings and your prompt. This starts an asynchronous generation and returns a generation_id. Required settings have no defaults.
  4. Poll GET /v1/videos/status/{generationId} until the generation completes, then fetch the result from content_url or call GET /v1/videos/content/{generationId}.

Preparation validates settings and does not start a generation or accept a prompt. The final charge is calculated for the generation request and may differ from the preliminary estimate.

Authentication

Video endpoints accept your API key as a Bearer token, X-Api-Key header, or apikey query parameter. Use the same account for model discovery, preparation, generation, status, and content requests.

Reference images and retries

A generation may include image references using owned file_id values. External reference URLs are not accepted. You can send an Idempotency-Key header when creating a generation to make retries of the same request idempotent.

Results and retention

Generations run asynchronously. The status response includes the state and, when complete, the content URL. Generated MP4 files are retained for 30 days.