Kling Models Guide
The Kling Unified Format covers 15 video models (kling-*); models differ only in capabilities and parameter ranges. All models go through the same two endpoints — Submit Task → Query & Download. This page introduces each model: what it does, which parameters it supports, and how to call it.
Asset models and routes are series-scoped: each series has its own asset models, and asset IDs are not interchangeable:
See Elements & Voices for details.
TIP
If you are using models named kling-video-* (the "Kling Format" that mirrors the vendor's official API structure), see the Kling Format docs. Model names and endpoints of the two lines are not interchangeable.
Model List
Of the 15 models, kling-2.5-turbo ships as a single version; the other seven come in two versions, kling-* and kling-v-*, with identical capabilities per row — the two models on each row share the same resolution, duration, audio and feature set:
Creating and referencing element / voice assets: see Elements & Voices.
Capability & Parameter Overview
Supported tiers per model (out-of-range submits return 400 with the legal values):
About audio: off silent (default, cheapest tier) / native generated voice-over / original keep source audio. Omitting the field means silent — pass it explicitly if you want sound.
Material channels (metadata.contents) and where they apply:
kling-2.5-turbo
Balanced entry workhorse for everyday text-to-video and image-to-video. No audio track — pick kling-2.6 or a v3 model if you need sound.
Image-to-video example (image present means i2v):
# Image-to-video: `image` present means i2v; Chinese prompt works best for the upstream model
curl -X POST https://api-model.shuntianchuang.cn/v1/videos \
-H "Authorization: Bearer sk-your-token" \
-H "Content-Type: application/json" \
-d '{
"model": "kling-2.5-turbo",
"prompt": "镜头缓缓推进,猫转头看向镜头",
"resolution": "1080p",
"image": "https://example.com/first-frame.jpg"
}'
import requests
resp = requests.post(
"https://api-model.shuntianchuang.cn/v1/videos",
headers={
"Authorization": "Bearer sk-your-token",
"Content-Type": "application/json",
},
json={
# Chinese prompt works best for the upstream model
"prompt": "镜头缓缓推进,猫转头看向镜头",
"model": "kling-2.5-turbo",
"resolution": "1080p",
"image": "https://example.com/first-frame.jpg",
},
)
task = resp.json()
print(task["id"]) # poll the query endpoint with this id
const resp = await fetch('https://api-model.shuntianchuang.cn/v1/videos', {
method: 'POST',
headers: {
Authorization: 'Bearer sk-your-token',
'Content-Type': 'application/json',
},
body: JSON.stringify({
model: 'kling-2.5-turbo',
// Chinese prompt works best for the upstream model
prompt: '镜头缓缓推进,猫转头看向镜头',
resolution: '1080p',
image: 'https://example.com/first-frame.jpg',
}),
});
const task = await resp.json();
console.log(task.id); // poll the query endpoint with this id
kling-2.6
Adds native audio: set metadata.audio to native to generate videos with sound (automatically generated and billed at 1080p). Also supports element / voice references — voice tasks are likewise forced to 1080p.
Talking image-to-video example:
# Talking i2v: audio=native is billed at 1080p; Chinese prompt works best for the upstream model
curl -X POST https://api-model.shuntianchuang.cn/v1/videos \
-H "Authorization: Bearer sk-your-token" \
-H "Content-Type: application/json" \
-d '{
"model": "kling-2.6",
"prompt": "镜头缓缓推进,海浪拍打礁石",
"resolution": "1080p",
"image": "https://example.com/first-frame.jpg",
"metadata": { "audio": "native" }
}'
import requests
resp = requests.post(
"https://api-model.shuntianchuang.cn/v1/videos",
headers={
"Authorization": "Bearer sk-your-token",
"Content-Type": "application/json",
},
json={
# Chinese prompt works best for the upstream model
"prompt": "镜头缓缓推进,海浪拍打礁石",
"model": "kling-2.6",
"resolution": "1080p",
"image": "https://example.com/first-frame.jpg",
"metadata": {"audio": "native"},
},
)
task = resp.json()
print(task["id"])
const resp = await fetch('https://api-model.shuntianchuang.cn/v1/videos', {
method: 'POST',
headers: {
Authorization: 'Bearer sk-your-token',
'Content-Type': 'application/json',
},
body: JSON.stringify({
model: 'kling-2.6',
// Chinese prompt works best for the upstream model
prompt: '镜头缓缓推进,海浪拍打礁石',
resolution: '1080p',
image: 'https://example.com/first-frame.jpg',
metadata: { audio: 'native' },
}),
});
const task = await resp.json();
console.log(task.id);
kling-3.0
Flagship quality: up to 4k, 3–15 s durations, multi-shot storyboarding, with original source audio.
Multi-shot example (customize scripting, see Submit Task · Multi-shot):
# Multi-shot storyboarding; Chinese prompts work best for the upstream model
curl -X POST https://api-model.shuntianchuang.cn/v1/videos \
-H "Authorization: Bearer sk-your-token" \
-H "Content-Type: application/json" \
-d '{
"model": "kling-3.0",
"prompt": "一只猫的一天",
"metadata": {
"multi_shot": true,
"shot_type": "customize",
"multi_prompt": [
{ "index": 1, "prompt": "清晨,猫在窗台伸懒腰", "duration": 3 },
{ "index": 2, "prompt": "午后,猫在草地上追蝴蝶", "duration": 4 }
]
}
}'
import requests
resp = requests.post(
"https://api-model.shuntianchuang.cn/v1/videos",
headers={
"Authorization": "Bearer sk-your-token",
"Content-Type": "application/json",
},
json={
# Chinese prompts work best for the upstream model
"prompt": "一只猫的一天",
"model": "kling-3.0",
"metadata": {
"multi_shot": True,
"shot_type": "customize",
"multi_prompt": [
{"index": 1, "prompt": "清晨,猫在窗台伸懒腰", "duration": 3},
{"index": 2, "prompt": "午后,猫在草地上追蝴蝶", "duration": 4},
],
},
},
)
task = resp.json()
print(task["id"])
const resp = await fetch('https://api-model.shuntianchuang.cn/v1/videos', {
method: 'POST',
headers: {
Authorization: 'Bearer sk-your-token',
'Content-Type': 'application/json',
},
body: JSON.stringify({
model: 'kling-3.0',
// Chinese prompts work best for the upstream model
prompt: '一只猫的一天',
metadata: {
multi_shot: true,
shot_type: 'customize',
multi_prompt: [
{ index: 1, prompt: '清晨,猫在窗台伸懒腰', duration: 3 },
{ index: 2, prompt: '午后,猫在草地上追蝴蝶', duration: 4 },
],
},
}),
});
const task = await resp.json();
console.log(task.id);
kling-3.0-turbo
The fast tier of v3: same 3–15 s long durations, faster and cheaper. No audio track.
Text-to-video example (defaults: 720p / 5 s / silent):
# Text-to-video with defaults (720p / 5 s / silent); Chinese prompt works best for the upstream model
curl -X POST https://api-model.shuntianchuang.cn/v1/videos \
-H "Authorization: Bearer sk-your-token" \
-H "Content-Type: application/json" \
-d '{
"model": "kling-3.0-turbo",
"prompt": "城市夜景延时摄影,车流光轨",
"duration": 5
}'
import requests
resp = requests.post(
"https://api-model.shuntianchuang.cn/v1/videos",
headers={
"Authorization": "Bearer sk-your-token",
"Content-Type": "application/json",
},
json={
# Chinese prompt works best for the upstream model
"prompt": "城市夜景延时摄影,车流光轨",
"model": "kling-3.0-turbo",
"duration": 5,
},
)
task = resp.json()
print(task["id"])
const resp = await fetch('https://api-model.shuntianchuang.cn/v1/videos', {
method: 'POST',
headers: {
Authorization: 'Bearer sk-your-token',
'Content-Type': 'application/json',
},
body: JSON.stringify({
model: 'kling-3.0-turbo',
// Chinese prompt works best for the upstream model
prompt: '城市夜景延时摄影,车流光轨',
duration: 5,
}),
});
const task = await resp.json();
console.log(task.id);
kling-3.0-omni
The all-round model: multi-shot, reference-video input, 4k and original audio are all supported.
Reference-video example (base_video carries the source video):
# Reference video via base_video; Chinese prompt works best for the upstream model
curl -X POST https://api-model.shuntianchuang.cn/v1/videos \
-H "Authorization: Bearer sk-your-token" \
-H "Content-Type: application/json" \
-d '{
"model": "kling-3.0-omni",
"prompt": "把视频画面转换成赛博朋克风格,保持人物动作不变",
"metadata": {
"contents": [
{ "type": "base_video", "url": "https://example.com/source.mp4" }
]
}
}'
import requests
resp = requests.post(
"https://api-model.shuntianchuang.cn/v1/videos",
headers={
"Authorization": "Bearer sk-your-token",
"Content-Type": "application/json",
},
json={
# Chinese prompt works best for the upstream model
"prompt": "把视频画面转换成赛博朋克风格,保持人物动作不变",
"model": "kling-3.0-omni",
"metadata": {
"contents": [
{"type": "base_video", "url": "https://example.com/source.mp4"}
]
},
},
)
task = resp.json()
print(task["id"])
const resp = await fetch('https://api-model.shuntianchuang.cn/v1/videos', {
method: 'POST',
headers: {
Authorization: 'Bearer sk-your-token',
'Content-Type': 'application/json',
},
body: JSON.stringify({
model: 'kling-3.0-omni',
// Chinese prompt works best for the upstream model
prompt: '把视频画面转换成赛博朋克风格,保持人物动作不变',
metadata: {
contents: [{ type: 'base_video', url: 'https://example.com/source.mp4' }],
},
}),
});
const task = await resp.json();
console.log(task.id);
kling-o1
Reference-video editing: upload a video and rewrite it per instructions. Without reference material only 5 / 10 s durations are available.
Reference-video editing example:
# Reference-video editing; Chinese prompt works best for the upstream model
curl -X POST https://api-model.shuntianchuang.cn/v1/videos \
-H "Authorization: Bearer sk-your-token" \
-H "Content-Type: application/json" \
-d '{
"model": "kling-o1",
"prompt": "去掉背景里的路人,整体调成暖色调",
"metadata": {
"contents": [
{ "type": "base_video", "url": "https://example.com/source.mp4" }
]
}
}'
import requests
resp = requests.post(
"https://api-model.shuntianchuang.cn/v1/videos",
headers={
"Authorization": "Bearer sk-your-token",
"Content-Type": "application/json",
},
json={
# Chinese prompt works best for the upstream model
"prompt": "去掉背景里的路人,整体调成暖色调",
"model": "kling-o1",
"metadata": {
"contents": [
{"type": "base_video", "url": "https://example.com/source.mp4"}
]
},
},
)
task = resp.json()
print(task["id"])
const resp = await fetch('https://api-model.shuntianchuang.cn/v1/videos', {
method: 'POST',
headers: {
Authorization: 'Bearer sk-your-token',
'Content-Type': 'application/json',
},
body: JSON.stringify({
model: 'kling-o1',
// Chinese prompt works best for the upstream model
prompt: '去掉背景里的路人,整体调成暖色调',
metadata: {
contents: [{ type: 'base_video', url: 'https://example.com/source.mp4' }],
},
}),
});
const task = await resp.json();
console.log(task.id);
kling-motion-3.0
Motion control: one character image plus one motion-reference video makes the character in the image follow the motions in the video. Material must be exactly one image + one video; the motion video must contain a clear human motion (a subject too small or missing an upper body is rejected and the task fails with the reason). The result duration follows the motion video — the requested duration is ignored. Accepts one element.
Motion-control example:
# Motion control: exactly one character image + one motion video; Chinese prompt works best
curl -X POST https://api-model.shuntianchuang.cn/v1/videos \
-H "Authorization: Bearer sk-your-token" \
-H "Content-Type: application/json" \
-d '{
"model": "kling-motion-3.0",
"metadata": {
"character_orientation": "video",
"audio": "original",
"contents": [
{ "type": "image", "url": "https://example.com/character.webp" },
{ "type": "video", "url": "https://example.com/dance.mp4" },
{ "type": "prompt", "text": "跟随视频中的舞蹈动作" }
]
}
}'
import requests
resp = requests.post(
"https://api-model.shuntianchuang.cn/v1/videos",
headers={
"Authorization": "Bearer sk-your-token",
"Content-Type": "application/json",
},
json={
"model": "kling-motion-3.0",
"metadata": {
"character_orientation": "video",
"audio": "original",
"contents": [
{"type": "image", "url": "https://example.com/character.webp"},
{"type": "video", "url": "https://example.com/dance.mp4"},
# Chinese prompt works best for the upstream model
{"type": "prompt", "text": "跟随视频中的舞蹈动作"},
],
},
},
)
task = resp.json()
print(task["id"])
const resp = await fetch('https://api-model.shuntianchuang.cn/v1/videos', {
method: 'POST',
headers: {
Authorization: 'Bearer sk-your-token',
'Content-Type': 'application/json',
},
body: JSON.stringify({
model: 'kling-motion-3.0',
metadata: {
character_orientation: 'video',
audio: 'original',
contents: [
{ type: 'image', url: 'https://example.com/character.webp' },
{ type: 'video', url: 'https://example.com/dance.mp4' },
// Chinese prompt works best for the upstream model
{ type: 'prompt', text: '跟随视频中的舞蹈动作' },
],
},
}),
});
const task = await resp.json();
console.log(task.id);
kling-motion-2.6
The previous generation of motion control: same rules as kling-motion-3.0 (exactly one image + one video, result duration follows the motion video) but no element references.
Calling it is identical to kling-motion-3.0 — just switch model to kling-motion-2.6. See Submit Task · Motion Control.
Next Steps