Two-minute AI music track cost: ElevenLabs per minute vs Sume flat
ElevenLabs Music lists $0.15 a minute: 2 minutes is $0.30. Sume charges a flat $0.125 per generation; Google lists $0.08 per Lyria 3.5 song. Fine print.

The cost of two minutes
ElevenLabs lists Music at $0.15 per minute on its API pricing page, so a two-minute track is 2 x $0.15 = $0.30. Sume charges a fixed $0.125 for one generation. Google's Gemini API pricing page lists Lyria 3.5 at $0.08 per song.
The shapes differ: one is billed by the minute, the other by the track. That changes the comparison as soon as your tracks get longer or shorter than two minutes.
Cost by number of tracks
The table assumes each track is two minutes. Sume's flat rate applies per generation, so it holds only if the model delivers a track of that length from your prompt.
| Tracks | ElevenLabs Music at $0.15 a minute | Sume at $0.125 each | Google Lyria 3.5 at $0.08 each |
|---|---|---|---|
| 1 | $0.30 | $0.125 | $0.08 |
| 5 | $1.50 | $0.625 | $0.40 |
| 20 | $6.00 | $2.50 | $1.60 |
How length works on Sume
The music router docs say there is no duration field: duration and duration_seconds are rejected. You ask for length in the prompt, for example "a 2-minute track", or with section markers like [0:00-0:30]. The price does not change with the length you write, but the engine decides what it returns, so check the file's real duration before you build a video around it.
The model defaults to sume/music-auto; you can choose lyria-3.5 or lyria-3-pro. After the job completes, request.routed_model tells you which engine ran.
Choosing on more than price
Google's rate is the lowest because it is the source of the Lyria models; Sume's $0.125 is the price of using them through the same jobs API as its other tools, with Idempotency-Key retries and signed webhooks. ElevenLabs' per-minute price may be better for a long track, and worse for many short ones.
- Under two minutes per track, a flat price is cheaper than a per-minute one.
- For long beds, test how long a track the model actually gives you.
- Prices change; re-check the vendor pages on the day you buy.
- A music price does not include licensing review for how you publish the track.
A budget for a video series
Say you publish 12 videos a quarter and make two candidate tracks per video, then keep one. That is 24 generations. On Sume that is 24 x $0.125 = $3.00. On ElevenLabs Music at $0.15 a minute, 24 two-minute tracks are 24 x $0.30 = $7.20. On Google's Lyria 3.5 price of $0.08 a song it is 24 x $0.08 = $1.92.
None of these figures includes the cost of listening, picking and mixing, which is usually the larger cost. Cheap tracks help only if you actually generate a few and choose; the habit of making two or three options is worth more than the difference between these rates.
Also note that the three prices measure different things: a minute of output, a song and a generation. Before you compare, write down what one unit means on each page, and re-read the page on the day you buy.
Sources
Related posts
More in Pricing
- Veo 3.1 4K at $0.30 to $0.60 a second vs Omni Flash 4K on Sume
Veo 3.1 Fast lists 4K at $0.30/s and Standard at $0.60/s; Lite has no 4K. Gemini Omni Flash 1.1 on Sume lists 4K at $0.30/s before the 1.25 multiplier.
- Veo 3.1 Lite $0.05-$0.08 per second vs Wan 3.0 on Sume
Gemini API lists Veo 3.1 Lite at $0.05/s for 720p and $0.08/s for 1080p. Sume does not carry Veo; here is the closest priced alternative at each resolution.
- Veo 3.1 Lite: 1,000 Flow credits vs per-second pricing on Sume
Google AI Pro's 1,000 Flow credits cover about 100 Veo 3.1 Lite videos. Here is the per-clip math, and how Sume's per-second SKUs compare.
- Veo 3.1 Lite at 720p: eight drafts cost the same as one Standard clip
At Google's listed rates, one 8-second Veo 3.1 Standard clip costs $3.20 and eight Lite 720p clips cost the same. Here is the draft-then-final math.
Written by Sume