Overview

Models

Two generation models, one per format. The endpoint you call selects the model.

Available models

Model	Produces	Credits	Output
`lullaby-v1` POST /api/v1/lullaby	A fully sung lullaby	1 (+1 with cover image)	MP3 audio, lyrics, title
`musical-v1` POST /api/v1/musical	A narrated bedtime story with a sung lullaby in the middle	2 (+1 with cover image)	MP3 audio, full narrated text, title

Model names are informational version labels. There is no model request parameter. Each endpoint always runs the current model for its format, and new versions will be announced here.

Generation time

Generation runs asynchronously and typically completes within a few minutes; musicals take longer than lullabies because they include both narration and song. Always poll GET /api/v1/jobs/{job_id} rather than assuming a fixed duration.

Choosing a model

Pick lullaby-v1 when you want a song: soothing, fully sung audio for wind-down playlists, baby monitors, or sleep apps.
Pick musical-v1 when you want a story: narrated long-form audio with a sung lullaby embedded, closer to a bedtime audiobook.
Set include_image: true on either model when you also need cover art (album tiles, story cards). It adds one credit and some generation time, so leave it off (the default) for audio-only flows.