Timeline render: 60 seconds costs $0.10, 61 seconds costs $0.20

A Sume timeline render of 60 seconds bills one minute, $0.10. A render of 61 seconds bills two minutes, $0.20. The public rate is $0.10 per ceil(output minute), the reserve is ceil(audio.duration_seconds / 60) minutes, and the job uses no provider inference, only worker ffmpeg. So the render line is small, but it steps up on the minute.

The step table

Duration is rounded up to whole minutes, with a floor of one minute and a ceiling of 30 minutes (audio.duration_seconds 1 to 1800).

What the step costs at volume

Two hundred 61-second videos cost 200 x $0.20 = $40.00 to render. Trim each to 60 seconds and the same two hundred cost 200 x $0.10 = $20.00. You save $20.00 by removing one second from each voiceover, and no provider call is involved.

The reverse matters for plans: a 10-second clip still bills the full minute, so batching ten 10-second spots into one 100-second timeline (2 minutes, $0.20) is cheaper than ten renders (10 x $0.10 = $1.00), if the delivery format allows one file.

A week of seven videos

Take a week of seven videos with lengths of 45, 58, 60, 61, 75, 119 and 121 seconds. They bill 1, 1, 1, 2, 2, 2 and 3 minutes, which is 12 minutes or $1.20. Trim the 61-second video to 60 and the 121-second video to 120 and the week bills 10 minutes, or $1.00.

That is a $0.20 saving across seven videos, so it will not change a budget on its own. It matters when the volume is high or when the output is long: a 30-minute render is the maximum and costs 30 x $0.10 = $3.00, while a 31-minute script cannot be rendered in one job at all and has to be split.

Also remember that a 10-second output still bills a full minute. If your product is a stream of very short clips, the render line is the one place where short output is not cheap per second: 10 s costs $0.10, which is $0.01 per second, against $0.00167 per second at 60 s.

How to apply it

Set audio.duration_seconds from your script, not from the rendered output, and check it against the minute boundary before you submit. If the script lands at 61 to 63 seconds, cutting a pause is cheaper than paying for a second minute. Read the timeline doc for the reserve rule and the 12-slot chunking behavior.

Where to enforce it

Put the minute boundary in your pipeline, not in people's heads.

• Compute ceil(seconds / 60) in code and show it next to the script length.
• Warn when the length is within 3 seconds above a boundary, such as 61 to 63 seconds.
• Log billed minutes against the billable_minutes field in the render result so that you catch drift.