Nine music variants for one Reel: a flat $0.125 each, $1.125 total

Music on Sume is a flat $0.125 per generation, so nine track variants cost $1.125 whatever the length. A cost table, the request body and what to check.

4 min readSume
All posts

Nine music variants cost $1.125 on Sume: the Music Router charges a flat $0.125 per generation on every music model, with the default engine currently Lyria 3.5. The price does not change with the length of the track you ask for.

That makes music the cheapest line in a Reel budget. The expensive part of the loop is the listening time, so decide the variant count by how many tracks you will actually audition.

Cost by number of variants

Total = variants x $0.125. Nothing else scales: the flat rate applies per generation, not per second.

Music Router, flat per generation (read 2026-10-09)
VariantsTotal
3$0.375
6$0.75
9$1.125
12$1.50
20$2.50

Ask for variants that differ

Docs say the duration is controlled in the prompt, not in a field: Sume rejects duration and duration_seconds. Vary tempo, key, or lead instrument between requests so the nine tracks are really nine options. Use one idempotency key per variant.

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: reel-42-music-v3" \
  -d '{"model":"sume/music-auto",
       "prompt":"Upbeat funk, 108 BPM, D minor, slap bass lead. A 20-second track. Instrumental, no vocals."}'

How to compare against the video line

At Wan 3.0 480p the 2-second minimum clip costs $0.125, so nine tracks ($1.125) cost about the same as 9 of those. A single 10-second 720p Wan 3.0 clip ($1.25) pays for ten music generations.

Pick one of nine without re-paying

Generate all nine, audition them in the dashboard or from the job results, and keep the job ids of the rejects so you can read the ledger later. Because the rate is flat per generation, regenerating a rejected variant costs another $0.125 and never changes the price of the others.

A useful rule: cap music at the cost of one cheap video clip. With a 10-second Wan 3.0 clip at 720p costing $1.25, ten music generations ($1.25) fit inside a single clip's price. If the audio decision is keeping you past that, the brief is the problem, not the budget.

Gotchas

  • negative_prompt is not supported: put exclusions in the positive prompt.
  • image_url can condition a track on a still, and the price stays flat.
  • sume/music-auto picks the engine; pin lyria-3.5 or lyria-3-pro from GET /v1/music-router/models if you need a specific one.
  • Check licensing and platform rules for the track before publishing; Sume docs describe generation, not the rights on a given platform.

Sources

Related posts

More in Use cases

All Use cases posts

Written by Sume