Skip to main content
POST
  • Generate full songs from a text description or custom lyrics (2 candidates per request)
  • Asynchronous task; poll by task ID
  • custom determines the meaning of prompt: custom lyrics vs inspiration description

Authorizations

string
required
Bearer Token. Get your API Key: API Key management

Body

string
required
Fixed value "suno"
boolean
Mode switch, default false
  • false: inspiration mode — gpt_description (or prompt) describes the music
  • true: custom lyrics mode — prompt is the lyrics text; title/style/negative_tags apply
string
Music description in inspiration mode, e.g. "lo-fi jazz for a rainy café". Not used in custom mode.
string
Lyrics text in custom mode (supports [Verse]/[Chorus] structure tags)
string
Song title (custom mode)
string
Style tags (custom mode), e.g. "dreamy synthwave, female vocal"
string
Elements to avoid (custom mode)
boolean
Generate instrumental-only music
string
Engine version: v6 / v6-wild / v6-mini
boolean
Max quality tier, billed at 2× ($0.10). Requires custom=true
string
Reuse a created vocal persona. Mutually exclusive with custom_model_id
string
Use a custom-trained model. Mutually exclusive with version and persona_id

Optional tuning parameters

style_weight, weirdness, audio_weight, vocal_gender, auto_lyrics, variety are all optional; upstream defaults apply when omitted.

Response

string
Task ID (tsk_aud_ prefix); poll via music task status
string
Initial status: submitted / queued / in_progress
On completion, result.music[] returns 2 candidates, each with audio_url, image_url, title, tags, lyrics, duration. Media URLs are mirrored to our CDN.