Kling 3.0 Motion Control API: 7.2 s driver video bills 8 s, $1.26

Sume lists Kling 3.0 Motion Control at $0.1575 per output second, billed on the ceiling of the driving video's length, from 1 to 30 seconds. A 7.2-second driver video rounds up to 8 seconds, so the job costs 8 x $0.1575 = $1.26. Call POST /v1/kling/3.0/motion-control with a still image and a motion video.

What it does

Motion Control animates a still image, or a ready avatar still, with the motion of a driving video you supply. The output length follows the driving video. The public model id is kling/3.0/motion-control; POST /v1/avatar-1.0/motion-control is an alias with the same body.

Cost table

The price description says Motion Control is priced per ceil(motion video second) after margin, with a duration of 1 to 30 seconds. Rate $0.1575 comes from the Sume catalog (read 2026-10-08).

Example

Set duration_seconds to the billed length (8 for the 7.2-second driver). character_orientation chooses whether the character faces as in the video or as in the image.

curl -X POST https://api.sume.com/v1/kling/3.0/motion-control \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: motion-001" \
  -d '{
    "image_url": "https://example.com/character.png",
    "motion_video_url": "https://example.com/dance.mp4",
    "duration_seconds": 8,
    "character_orientation": "video",
    "keep_original_sound": false
  }'

Preparing the driving video

Output length follows the driving video, so cut the motion reference to the length you want before you submit. Each second you trim saves $0.1575. Cutting a 12-second reference to 8 seconds takes the job from $1.89 to $1.26, a saving of $0.63. Use video trim at $0.02 per job to cut it, which pays for itself whenever it removes more than one second.

Pass keep_original_sound: true only if you want the driving video's audio on the result. A character image with a clear, front-facing subject usually gives the model the most to work with, but the docs make no promise about quality, so test with one short job first.

Budgeting a batch

Ten characters animated by the same 8-second driver cost 10 x $1.26 = $12.60. Because price is linear in billed seconds, halving the driver to 4 seconds halves the bill to $6.30. A 30-second driver is $4.725 per character. Set a spend plan before the batch, and use one short test job to check that the orientation setting (video or image) gives the pose you want, since a wrong setting costs the full job.

What Sume does not list

Kling 4.0 Flash is in limited early access per press coverage, and the Kling pricing page did not show Kling 4.0 API pricing when fetched on 2026-10-08, so there is no public API price to compare. Sume lists Kling 3.0 Motion Control only. Poll GET /v1/jobs/:id/status and /result for the video.