Music for 20 video ads, three takes each: 60 generations, $7.50

The short answer

Sume charges a fixed $0.125 for each accepted music generation. Three takes for each of 20 video ads is 20 x 3 = 60 generations, and 60 x $0.125 = $7.50. The docs state that the price does not change with prompt length or optional image conditioning, so a long, detailed prompt costs the same as a one-line prompt.

Cost by number of takes

The per-generation price is the same on Music 1.0 and on every Music Router model: the Music Router page says each model charges the fixed Music price, and lists the provider price only for reference.

The catalog list vs what you pay

The Music Router catalog shows a provider list price for each engine (for example lyria-3.5 and lyria-3-pro) for reference only. The Music Router page is explicit that every model charges the fixed Music price per audio generation, so choosing a model with a lower list price in the catalog is not a way to lower your bill. Use sume/music-auto or pin an engine for sound and length reasons, not price.

Because billing is per accepted generation, a rejected request does not appear in the total above. The docs describe the price as applying to accepted generations; check the job status of a failed request before you assume it was charged.

Sizing takes for a bigger campaign

The same fixed price makes scaling simple. Fifty ads at three takes each is 150 generations and $18.75. One hundred ads at two takes each is 200 generations and $25.00. If you generate one track and reuse it across five video variants, 100 variants need only 20 tracks, or $2.50.

Because the price ignores prompt length, put the detail in the prompt: tempo, key, instruments, mood and the exclusions you want, as the Music Router example prompt does. An extra sentence costs nothing, and a better first take saves a $0.125 retry.

Fitting music into a whole ad budget

At 12.5 cents a take, music is a small line. In a 30-second vertical ad assembled with Timeline 1.0 ($0.10 per rounded-up output minute), three music takes are 37.5 cents, the Timeline render is 10 cents, and the three 10-second clips at 720p on Wan 3.0 are $3.75. Music is under 9 percent of that $4.22 total.

See Music 1.0 and Music Router for the request fields, and Timeline 1.0 for assembly pricing. Music 1.0 is a compatibility route that resolves through the Music Router, so new integrations use the router.