JUN 17, 2026

Kling 3.0 Turbo on the API

Kling API v1 now supports Kling v3 Turbo (kling-v3-0-turbo) — a faster variant of Kling 3.0 that keeps v3.0 quality while cutting generation time. It’s available on POST /videos/text2video and POST /videos/image2video-frames, with std/pro modes and 3–15s durations. Audio is always on, and Turbo leans into multi-shot output — so a single prompt can carry spoken dialogue and scene changes without any extra parameters.

This post walks one image-to-video demo end to end: a single noir portrait becomes a 10-second 1080p clip with two spoken lines and three settings — all produced through the public Kling API v1.

The demo: one portrait → a 10s noir clip

Source image

A single 9:16 portrait, uploaded via POST /assets and passed as the start frame. The output aspect ratio is derived from the image (no aspect_ratio parameter on v3 image-to-video).

Noir woman in a red hat and coat — start frame
Start frame — image
(uploaded via POST /assets)

Generate the clip

One call to POST /videos/image2video-frames. model_name: kling-v3-0-turbo, mode: pro (1080p), duration: 10. The prompt does all the directing — two spoken lines for the always-on audio, and an office → alley → rooftop progression that Turbo renders as scene cuts within the single clip. No aspect_ratio, no enable_audio, no multi-shot parameters.

curl --location 'https://api.useapi.net/v1/kling/videos/image2video-frames' \
--header 'Authorization: Bearer user:12345-…' \
--form 'email="[email protected]"' \
--form 'model_name="kling-v3-0-turbo"' \
--form 'mode="pro"' \
--form 'duration="10"' \
--form 'image="https://s15-kling.klingai.com/…"' \
--form 'prompt="The woman in the red hat leans toward the camera in a dim, rain-streaked office and says in a low, confident voice: Everyone'\''s chasing the diamond — but only I know where it'\''s hidden. She turns and strides into a neon-soaked alley; the scene then cuts to a glittering rooftop overlooking the night skyline, where she glances back over her shoulder with a sly smile and adds: Catch me if you can. The camera sweeps up from the street to the city lights."'

Result

From one still portrait: two spoken lines on an in-character voice, and a clean cut from the rain-streaked office to the neon alley to the rooftop skyline — all driven by the prompt alone, on the faster Turbo model.

Turbo or standard 3.0?

Turbo is the faster option and was a great fit for this image-to-video demo. When you need the full v3.0 feature set, use kling-v3-0 (standard 3.0).

Model capabilities change with each Kling release, so for the current, authoritative comparison check the live Model Matrix and the text2video and image2video-frames docs.

Get started

Kling API v1 docs · Setup Kling · POST /videos/text2video · POST /videos/image2video-frames