Kling motion control 23-second clip: $3.62, and 22.4 s bills as 23
Kling 3.0 Motion Control on Sume is $0.1575 per output second: a 23-second driving video is 23 x 0.1575 = $3.62, and a 22.4 s one is rounded up to 23.

A 23-second Kling 3.0 Motion Control clip on Sume costs 23 x $0.1575 = $3.6225, which shows as $3.62 on a cent-rounded balance (as of 2026-10-09). A driving video of 22.4 seconds is billed as 23, because the catalog prices motion control per ceil of the motion-video second.
Motion control takes a still image and a driving video, and returns the still moving the way the video moves. The output is as long as the driving video, so the length of your reference clip is the whole cost question.
Rounding up, with numbers
The catalog's pricing basis says Sume reserves ceil(duration_seconds) times the per-second list rate after margin. The request body asks for duration_seconds between 1 and 30, and the driving video itself may be at most 30 seconds. If your reference is 22.4 seconds long, send 23. The table shows the per-length cost for a few reference durations that people actually cut.
| Driving video | Billed seconds | Arithmetic | Cost |
|---|---|---|---|
| 22.4 s | 23 | 23 x 0.1575 | $3.62 |
| 23.0 s | 23 | 23 x 0.1575 | $3.62 |
| 23.1 s | 24 | 24 x 0.1575 | $3.78 |
| 30.0 s | 30 | 30 x 0.1575 | $4.73 |
Why trimming by 0.1 second matters
The step from 23.0 to 23.1 seconds costs $0.16, which is the price of one more second. For one clip that is small. For a library of 200 reference moves it is $32 if each one overshoots a whole second by a tenth. If you control the reference clips, cut them on a whole second. The docs list a video trim route at $0.02 per job, so trimming an overshoot costs less than the second it saves.
What else the request controls
Besides the still and the motion video, the body accepts keep_original_sound (default true, so the driving clip's audio stays in the output) and character_orientation (video by default, or image to keep the still's framing). An optional prompt of up to 2,000 characters only steers appearance, since motion and framing follow the video.
Send exactly one visual source: an image_url or an avatar id or handle. The motion video must be a fetchable public HTTPS URL.
- Price: $0.1575 per output second, 1-30 seconds.
- Billed unit: ceil of the motion-video second.
- Submit-time reservation is refunded if the job fails before capture.
Where the limit bites
There is no way to stretch a 30-second cap. For a longer move, split the reference into pieces and join them afterwards. A 23-second move is comfortably inside one job, so there is nothing to split. See the four routes that share one stored model id if you are choosing between the URLs, and the 30-second price breakdown for the ceiling.
A batch example
A team that animates one mascot with 15 different 23-second moves pays 15 x $3.6225 = $54.34 at the catalog rate. If five of those references are really 22.0 seconds, they bill as 22 seconds and cost $3.465 each, saving 5 x $0.1575 = $0.79 over the 23-second figure. The savings are small per clip, but the habit of declaring duration_seconds from a measured value, instead of a round guess like 30, avoids the large error: declaring 30 for a 23-second clip reserves $4.725 at submit, $1.10 more than the clip needs.
Sources
Related posts
More in Media tools
- Kling Motion Control on a 7.2-second driving clip bills 8 s: $1.26
A 7.2-second driving video is billed as 8 seconds on Sume's Kling 3.0 Motion Control: 8 x $0.1575 = $1.26. Trim to 7.0 s to pay $1.1025 instead.
- Six Motion Control jobs: 29.8 s of footage bills 32 s, $5.04
Six Kling 3.0 Motion Control jobs of 4.5, 6.2, 3.8, 5, 7.4 and 2.9 seconds bill 5+7+4+5+8+3 = 32 seconds on Sume: 32 x $0.1575 = $5.04.
- Does Kling motion control follow my prompt? Only for appearance
On Sume's Kling 3.0 Motion Control the optional prompt (up to 2,000 characters) steers appearance only; motion and framing follow the driving video.
- Korean captions: slam returns 400, so use korean-ad with language ko
Latin caption styles have no Hangul glyphs, so Korean text on slam, punch or tiktok-green returns 400 caption_hangul_text_latin_style. Use korean-ad.
Written by Sume