Descript per-second smoothing credits vs flat API job prices
Descript now bills smoothing by the second. Here is how that compares with Sume media jobs, which charge a flat amount per job or per output minute.
Descript's September 17, 2026 changelog roundup says smoothing is priced by the second now: 0.5 AI credits per second of regenerated audio and 1.0 per second of video, instead of a flat charge per cut. So a long cut costs more than a short one. Sume's media jobs take the opposite approach: most are a flat price per job, and the one per-duration price, Timeline render, is $0.10 per output minute. Neither is cheaper in general; it depends on how long your cuts are.
The numbers, dated
Descript reports credits, not dollars, so this post does not convert them: the dollar value of a credit depends on your Descript plan. Sume's prices are in dollars and come from the docs pages named in the table.
| Operation | Descript smoothing (changelog) | Sume job |
|---|---|---|
| Smoothing a cut | 0.5 credits/s audio, 1.0 credits/s video | No direct equivalent; trim is a plain cut |
| Cut a clip | Part of editing | video-trim $0.02 per job |
| Pull the audio | Part of editing | audio-detach $0.01 per job |
| Transcribe | Plan dependent | video-inspect transcribe:true $0.01 per audio minute |
| Assemble and render | Plan dependent | timeline-1.0 $0.10 per ceil output minute |
| Source | Descript changelog | Video trim, Timeline |
Working out a break-even
Take a 30-second cut. Descript's rule means smoothing that much video regenerates 30 s at 1.0 credit per second, or 30 credits. Sume's trim of the same 30 s is $0.02, but trim does not regenerate or smooth anything: it cuts frames. So the comparison only holds when what you need is the cut itself. If you need a regenerated, smoothed join, there is no flat Sume job for that, and you would use a Timeline transition (fade, dissolve, wipe up to 1 s) between two real clips instead.
The pricing shape is the real difference. A per-second charge scales with every second you regenerate. A flat per-job charge makes a batch of 100 clips easy to forecast: 100 trims are $2.00, and a plan call for a timeline is unbilled.
How to compare for your own work
Count the operations in one finished video: how many cuts, how many seconds regenerated, how many minutes rendered. Multiply by each side's published rate, using your own plan's credit value for Descript. Do it for your shortest and your longest typical video, because per-second pricing and per-job pricing diverge most at the extremes.
Limits
This compares published rates only; it measures no quality. Descript's page is a changelog, so its numbers can change at any time, and the label above says when it was read. Sume's rates should be confirmed live in GET /v1/catalog before you budget on them.
Sources
Related posts
More in Comparisons
- Dreamina Long Video Mode: 3 minutes vs Sume's 30 s Seedance
Dreamina says its Seedance 2.5 Long Video Mode makes videos up to three minutes. Sume's seedance-2.5 makes 4-30 s per request, so long films are joined clips.
- ElevenLabs at a $22B valuation: what an audio API buyer should check
A $300M employee tender valued ElevenLabs at $22B, double February's $11B. What a valuation does and does not tell someone building on its audio API.
- ElevenLabs Avatars wants 3 to 5 reference images: what Sume needs
ElevenLabs recommends 3-5 reference images from different angles for Avatars. Sume Avatar 1.0 builds an avatar handle from one photo, a prompt, or simple props.
- ElevenLabs Avatars has no API at launch: the Sume Avatar API
ElevenLabs says an Avatars API is planned but not live. If you need to automate avatar videos now, here is the Sume Avatar 1.0 route and what it needs.
Written by Sume