Provider Kling

Kling

Modes: video · image · audio · Models: 13

Vendor: Kling (Kuaishou) · Official API docs: Kling API

Kling (by Kuaishou) is a video-first model family with text-to-video, image-to-video, start/end-frame control, motion control, talking avatars, and native audio. Kling V3 is the flagship: the only version with multi-shot storyboards and native 4K (3–15s). The lineup also spans image generation and a TTS / video-to-audio side.

Models

id	Name	Input type
kling-v3	Kling V3	t2v
kling-v3-omni	Kling V3 Omni	t2v
kling-v2-6	Kling V2.6	t2v
kling-video-o1	Kling Video O1	t2v
kling-video-effects	Kling Video Effects	i2v
kling-motion-control-v3	Kling Motion Control V3	i2v
kling-motion-control	Kling Motion Control 2.6	i2v
kling-avatar	Kling Avatar	i2v
kling-3.0-image	Kling 3.0 Image	t2i
kling-o1-image	Kling O1 Image	t2i
kling-v2-new-image	Kling V2 New Image	t2i
kling-t2a	Kling T2A	t2a
kling-v2a	Kling V2A	v2a

CLI

# text-to-video with native audio
gen-ai generate -m kling-v3 \
  -p "a designer opening Picsart on a sunlit desk, cinematic, shallow depth of field" \
  --ar 16:9 -d 5 --audio-gen

# image-to-video from a start frame
gen-ai generate -m kling-v3 -p "camera slowly pushes in, she smiles" --start-frame ./hero.jpg

# image generation
gen-ai generate -m kling-3.0-image -p "a neon koi pond at dusk, ultra detailed" --ar 1:1

# text-to-audio
gen-ai generate -m kling-t2a -p "warm narrator voice reading a product tagline"

MCP

{ "name": "picsart_generate",
  "arguments": {
    "model": "kling-v3",
    "prompt": "a designer opening Picsart on a sunlit desk, cinematic",
    "aspectRatio": "16:9",
    "duration": 5,
    "generateAudio": true
  } }

{ "name": "picsart_generate",
  "arguments": {
    "model": "kling-3.0-image",
    "prompt": "a neon koi pond at dusk, ultra detailed",
    "aspectRatio": "1:1"
  } }

Parameters — kling-v3

Param	CLI flag	Type	Values
prompt	-p	text	required (≤2500 chars)
aspectRatio	--ar	enum	16:9 · 9:16 · 1:1 (default 16:9)
duration	-d	enum	3 · 5 · 8 · 10 · 12 · 15 (seconds, default 5)
negativePrompt	--neg	text	discourage unwanted content
generateAudio	--audio-gen	boolean	native audio track (default true)
startFrame	--start-frame	file	first-frame keyframe (image)
endFrame	--end-frame	file	last-frame keyframe (image)
multiShot	--multi-shot	boolean	multi-scene storyboard mode (default false)
shotType	--shot-type	enum	customize · intelligence (default customize)
multiPrompt	--multi-prompt	object[]	up to 6 storyboard shots {index, prompt, duration}
voiceList	--voice-list	object[]	up to 2 voice references {voice_id}
elementList	--element-list	object[]	up to 3 element references {element_id}
staticMask	--static-mask	file	static-region motion-brush mask (image)
renderingSpeed	--rendering-speed	enum	std (Standard) · pro (Pro) · 4k (4K) (default std)

Source: gen-ai models info kling-v3 --json. Multi-shot is incompatible with start/end-frame images; native 4K (renderingSpeed: 4k) is V3-only.

Pricing

gen-ai pricing kling-v3 -d 5 --rendering-speed pro

Cost scales with duration, rendering speed / resolution (std 720p · pro 1080p · 4k), and audio.