What $5.00 buys on Sume: seconds, characters, minutes and jobs by rate

Five dollars buys 27 seconds of standard avatar video, 26 seconds of Fabric at 720p, 105,263 characters of text to speech, 500 minutes of transcript, 222 background removals or 40 music tracks. The table divides $5.00 by each published Sume rate and rounds down to whole units.

The rates and the division

Rates come from the API pricing page and the guides linked from the Models overview. No margin or minimum is applied beyond what the rate states, and every figure is a floor division, so the last unit is only counted if it is fully covered.

Caveats before you plan around it

Per-second products bill per whole second, rounded up, and some have an input window: Fabric takes up to 300 seconds, H3 Max lip-sync 5 to 14.8 seconds, Kling Motion Control 1 to 30 seconds. Text to speech has a 1 cent minimum per job, so $5.00 of 100-character lines buys at most 500 jobs, far fewer than the 105,263 characters the rate would suggest.

The flat-job tools show the largest ratio between rate and use. A trim costs $0.02 and a caption job $0.20, so a pipeline of one trim, one caption job and one render minute on a short clip is $0.32 and $5.00 covers 15 of them.

Using the table

Use it to sanity-check a quote. If a job is expected to cost $0.30 and the table says a 4 second standard avatar clip costs $0.736, the quote is for something else. For generation families with variable prices, such as Image 1.0 and Video 1.0, use the GET /v1/catalog estimate and band rather than this table.