Kling 3 mute plus a Music 1.0 bed: $1.625 vs $2.10 with sound

On Sume a 10-second Kling 3 clip costs $2.10 with native audio and $1.40 muted, a difference of 70 cents (7 cents a second). If you would rather choose the sound yourself, a muted clip plus one Music 1.0 bed ($0.125) plus a Timeline render ($0.10) comes to $1.625, which is 47.5 cents under the clip with sound. Whether that is worth it depends on how much control of the audio you need.

The numbers

Prices are billable (catalog list times 1.25, rounded up to the cent) and were computed from the repository catalog as of 2026-10-08. Kling 3 accepts 4 to 15 seconds at 720p and 1080p in 16:9, 9:16 and 1:1, and its audio and mute prices are the same at both resolutions.

Arithmetic

Audio: 210 cents. Mute: 140 cents. Audio surcharge: 70 cents over 10 s, or 7 cents a second.

Mute path: 1.40 + 0.125 + 0.10 = 1.625 dollars. Savings: 2.10 - 1.625 = 0.475, so 47.5 cents.

What the cheaper path gives up

Music 1.0 has no duration parameter, and its docs say a duration or duration_seconds field is rejected. You ask for length in the prompt (the docs give "a 2-minute track" as an example), so measure the result. A bed for a 10-second clip is far shorter than the structured songs it is built for; you take the first 10 seconds with audio.source_in and audio.duration_seconds set to 10.

The Timeline path gives up the sound that the model makes for the picture. A music bed is only a bed, so if the scene needs sounds that match the action, the 70 cents may be the better buy.

The video API exposes a generate_audio boolean, and its default is the audio capability of the model. Send false to get the mute price. Confirm the returned clip with a probe before you build on it.

When to use which

Many clips share one bed: the Music cost spreads out. Twenty ten-second mute clips plus one shared bed cost 20 x $1.40 + $0.125 = $28.125 before renders, against 20 x $2.10 = $42.00 with native audio.

One clip, one scene, sound matters: pay the 70 cents.

Where the money goes

The sound-on clip is $2.10 for 10 seconds and the muted clip is $1.40, so native sound costs 210 - 140 = 70 cents. The muted route adds a Music 1.0 generation at a fixed $0.125 and one Timeline render at $0.10 per started output minute, which gives 1.40 + 0.125 + 0.10 = $1.625 before the platform rounds the line items you actually see. The sound-on route needs no Timeline step if you do not need to mix anything, so its total stays $2.10.

That means the muted route is only cheaper when the gap between the two totals is positive: 2.10 - 1.625 = $0.475 per clip. Across 20 clips that is $9.50. It stops being cheaper as soon as the bed needs a second generation, because two Music 1.0 generations are 2 x 0.125 = $0.25 and the total becomes 1.40 + 0.25 + 0.10 = $1.75.

What you give up

Native sound is generated with the picture, so it follows the scene. A bed from Music 1.0 follows only its own prompt, and the Music docs say the prompt controls the length and that the result should be examined, because a prompt axis is a direction and not a guaranteed output. If the clip is a dialogue scene, the muted route removes the speech you wanted. If the clip is a product shot that needs a bed under a voice-over, it is a good fit.

Check the length before you pay for the Timeline step. The bed must cover the whole 10 seconds; if the track runs longer, the Timeline audio spine is what sets the output length, so write the length into the music prompt rather than relying on a trim.