Music generation
Create music from a prompt and supported model settings.
- Open a chat and select Audio.
- Choose a music model such as Music Pro or Music 3 from the available options.
- Describe the style, instrumentation, mood, and intended use.
- Add lyrics or choose duration only when the selected model supports those controls.
- Review the estimate and generate. Play the result and download the version you want to keep.
Example prompt
An understated instrumental bed for a product film: warm electric piano, soft percussion, a steady pulse, and a restrained ending. Leave room for narration.
For vocal music, provide lyrics in the supported lyrics field and use the prompt for musical direction. Model capabilities vary, so a setting available on one model may be absent on another.
Iteration and cost
Change one aspect at a time: instrumentation, tempo description, vocal delivery, or duration. Every generation spends credits; playback of an existing output does not regenerate it.
Music uses the same balance as images, speech, video, and the API. It is not billed from the separate per-minute dollar balance described in the old documentation.
See music and sound through the API for integration examples.