Suno API: the music API options and what Sume offers

Sume does not offer a Suno API: its docs list no Suno model. What it does offer is a prompt-to-audio endpoint, POST /v1/music-router/generate, that runs Lyria and returns a Sume-hosted audio file. If you specifically need Suno v6, announced on Suno's blog on 2026-09-09, take that up with Suno; if you need music from an API, the shape below is what Sume documents.

What does a Sume music request look like?

A JSON body with a prompt of 1 to 5000 characters. model is optional and defaults to sume/music-auto; the routable ids are sume/music-auto, lyria-3.5 and lyria-3-pro. duration and duration_seconds are rejected, so length is steered in the prompt.

Send an Idempotency-Key header, as the docs' example does.

curl -X POST https://api.sume.com/v1/music-router/generate \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: music-router-001" \
  -d '{
    "model": "sume/music-auto",
    "prompt": "Warm lo-fi hip hop, 84 BPM, C minor. A 30-second track. Instrumental, no vocals."
  }'

How do I get the audio back?

The call creates a job. Poll GET /v1/jobs/{id}/status and then read GET /v1/jobs/{id}/result; the audio is the entry in result.artifacts[] whose type is audio, typically audio/mpeg on media.sume.com. For push instead of poll, see webhook callbacks.

What does it cost and what is missing compared with Suno?

A fixed $0.125 per accepted generation on Music 1.0, and every router model charges the fixed Music price. What Sume's docs do not list is a Suno model or any Suno-specific feature, so nothing on this page describes Suno's own API or plans.

Should I use sync or async mode?

The mode field accepts async, sync, subscribe and webhook. For sync and subscribe you can set wait_timeout_seconds from 0 to 30. Music takes longer than a quick image call, so most integrations submit, then poll or use a webhook and keep the job id.

Add metadata if you want your own reference stored on the job; the docs say it is caller metadata stored on the job and not sent to the provider.

Which errors should my client handle?

Three from the docs. An unknown model fails with 400 model_not_found and a catalog_url. A non-empty negative_prompt returns HTTP 400 with public_reason=negative_prompt_unsupported, so put exclusions in the positive prompt, such as no vocals. And duration or duration_seconds are rejected. Handle these as request bugs to fix, not as transient failures to retry.

What is the short version?

For a Suno API, go to Suno. For prompt-to-audio through one HTTP call with a fixed price and Sume-hosted files, the Music Router endpoint is what Sume documents. Send a prompt, keep the job id, poll or use a webhook, and fetch the audio artifact. Remember length lives in the prompt, not in a field.