Why Sume Music is $0.125 a track: Lyria list plus 25 percent
Sume Music bills $0.125 per track, which is the provider list of $0.10 times 1.25. Read the catalog fields that show it, and what the fixed price covers.

Sume bills $0.125 per Music 1.0 or Music Router track because the catalog applies a 1.25 multiplier to the provider list price, and for Lyria 3.5 that list price is $0.10. The Music Router schema says so directly: provider list times 1.25 per audio on every Music Router model, with the list shown for reference.
The price is per generation, not per second or per character, which fits the product: the Music 1.0 docs state there is no duration field, so the bill cannot depend on length. Prices below were checked 2026-10-10 against the docs and the catalog schema.
Where can you see the numbers?
Each row in GET /v1/music-router/models has pricing.list_usd_micros_per_audio (the provider list, for reference) and pricing.billable_usd_micros_per_audio (what Sume charges), plus a billable_formula string. Micros are millionths of a dollar, so 100000 micros is $0.10 and 125000 micros is $0.125.
| Quantity | Micros | Dollars |
|---|---|---|
| Provider list, Lyria 3.5 | 100000 | $0.10 |
| Margin multiplier | 1.25 | - |
| Billable per track | 125000 | $0.125 |
What does the fixed price cover?
One accepted generation, whatever the prompt length up to 5,000 characters. It also covers an optional public HTTPS image_url. It does not cover repeated attempts: each new request is a new generation at the same price, and there is no seed to reproduce a track. A request rejected before generation, such as a duration field or a non-empty negative_prompt, returns a 400 and never reaches the model.
- Prompt of 100 characters or 5,000: same price.
- Two takes: $0.25.
- A 400 error for a rejected field is not a generation.
Do the three router ids differ in price?
No. sume/music-auto, lyria-3.5 and lyria-3-pro share the same fixed billable price, so choosing between them is about sound and repeatability, not cost. sume/music-auto resolves to lyria-3.5 today and may move without notice, so log job.request.routed_model for each track.
| Id | Routing | Price per track |
|---|---|---|
sume/music-auto | auto | $0.125 |
lyria-3.5 | pass through | $0.125 |
lyria-3-pro | pass through | $0.125 |
What does this mean for a budget?
Budget by takes, not seconds. If you keep one take in three, a kept track costs $0.375; add the fixed $0.01 for each timeline step when you join the music to a voiceover. For long soundtracks, the music docs and a stored page cover looping or concatenating, which adds a $0.01 timeline job rather than another generation.
Worked example for a month of music
A small studio makes 12 explainers a month and wants three candidate tracks per video, keeping one. That is 36 generations, so 36 x $0.125 = $4.50 a month, and each kept track effectively costs $0.375. If a policy rejection happens before generation it does not make a track; a rejected request that reaches no model is not a generation. Add the timeline join at $0.01 per video if you mix the music under narration: 12 x $0.01 = $0.12.
Compare that with a larger campaign of 200 tracks, which is $25.00 at the fixed price. The per-track number never changes with length or prompt size, which makes the budget easy to check against the invoice. If your invoice shows a different per-track amount, read the live catalog row first and compare it with the docs.
| Scenario | Generations | Cost |
|---|---|---|
| 12 videos, 3 candidates each | 36 | $4.50 |
| 12 joins under narration | 12 jobs | $0.12 |
| Campaign of 200 tracks | 200 | $25.00 |
Sources
Related posts
More in Pricing
- How Sume pricing works: plans, one wallet, published model rates
Sume plans set access and concurrency. Usage draws from one prepaid wallet at each model's published USD rate, for generation, the Agent, Formats, and the API.
- AI avatar video API pricing: cost per second and per minute
Sume bills AI avatar video per second by quality tier, with separate rates when you send a product image. Per-minute costs for standard, plus, and max.
- AI video generation cost per video: what one Sume API run cost
To see what one Sume run or video job cost, call GET /v1/usage with run_id or job_id and read debited_usd: the wallet deduction, agent turns included.
- Do failed AI video generations cost credits? Reserve, capture, refund
No. Sume reserves a job's estimated USD cost at submit, captures usage only on completion, and releases or refunds the hold if the job fails or is canceled.
Written by Sume