Kling 3.0 voice control is $0.196 a second on fal; Sume has two rates

On Sume, kling-3 has two prices: $0.14 a second with audio off and $0.21 with audio on. fal's Kling 3.0 Pro image-to-video page, read 2026-10-05, adds a third tier at $0.196 a second for voice control, and Sume does not list a separate voice-control price or field.

The three fal tiers

fal's page lists $0.112 per second with audio off, $0.168 with audio on, and $0.196 per second when voice control is used. It accepts 3 to 15 seconds, requires a start image with an optional end image, and takes the aspect ratio from the start image.

Sume's catalog row for kling-3 is 4 to 15 seconds at 720p or 1080p and accepts no reference_*_urls. It prices two states, from the Video Router docs rule of list times 1.25.

Same clip, three prices

The table prices a 10-second clip. The voice-control row is the fal list only, since Sume does not sell that tier.

Choosing on Sume

Audio on costs $0.70 more over 10 seconds, which is the 50% step the audio pricing post breaks down. If you need a specific voice, the practical routes are to generate silent Kling footage and add speech from a text-to-speech step, or to use a model with native audio.

Note the other limits: kling-3 needs a first frame on Sume as well, so you cannot start from text alone with it; see the source-image post for the rows that do.