Kling Video O3 4k

kwaivgi/kling-video/o3/4k/reference-to-video

Kling Video O3 4k by KwaiVGI - animate still images into dynamic videos with AI. Transform photos into cinematic clips with natural motion, camera movement, and optional audio generation.

Base price
$2.10USD / run
Execution
async
Model type
video
Input fields
9

Try the model

Playground

Open playground
Input
Text prompt for video generation. Either prompt or multi_prompt must be provided, but not both.

PNG, JPEG, WebP, or GIF · 20 MiB maximum

Image to use as the first frame of the video.

PNG, JPEG, WebP, or GIF · 20 MiB maximum each

Reference images for style/appearance. Reference in prompt as @Image1, @Image2, etc. Maximum 7 total (elements + reference images) is allowed.

PNG, JPEG, WebP, or GIF · 20 MiB maximum

Image to use as the last frame of the video.
Video duration in seconds (3-15s). Allowed values: 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15.
Whether to generate native audio for the video.
List of prompts for multi-shot video generation.
The type of multi-shot video generation. 'intelligent' lets the model automatically determine shot structure. Allowed values: customize, intelligent.
The aspect ratio of the generated image. Allowed values: 16:9, 9:16, 1:1.
OutputReady

Your output will appear here

Complete the inputs, then click Run.

Specifications

Pricing

Base price
$2.10 / run
Billing formula
params.duration * 0.42

Context & modalities

Input
Schema-defined
Output
video

Capabilities

Chat
Not supported
Vision
Not supported
Reasoning
Not supported
Structured output
Not supported
Function calling
Not supported
Audio input
Not supported

Access

Provider
KwaiVGI
Model ID
kwaivgi/kling-video/o3/4k/reference-to-video
Execution
async
API
Unified Run API
Endpoint
/v1/run

API README

Kling Video O3 4k

reference to video in the Kling Video lineup is built to build a moving scene around supplied visual references. The kling video o3 4k reference to video configuration combines that transformation with the quality and motion profile represented by this exact model version, so teams can choose it deliberately among adjacent family variants. It is useful when the input already establishes part of the creative intent and the model must supply a polished result rather than a generic media conversion, with subject identity, scene logic, visual hierarchy, and the delivery goal kept explicit.

For production work with kwaivgi/kling-video/o3/4k/reference-to-video, begin with the non-negotiable content, then describe the desired change, framing, action, atmosphere, and finishing cues in that order. Separate what must remain recognizable from what may vary, and prefer concrete nouns and observable actions over abstract praise. That structure makes outputs easier to compare across storyboard passes, campaign variants, catalog assets, and other repeatable creative pipelines.

Highlights

Reference identity retention. Carries recognizable characters, products, costumes, or design cues into motion.

Cross-reference composition. Brings multiple visual ingredients into one directed scene.

Narrative motion generation. Creates action, camera movement, and environment response around references.

kling video o3 4k reference to video asset continuity. Supports recurring visual identities across storyboard and campaign variations. This is the defining creative strength of the kling video o3 4k reference to video configuration.

Pricing

DurationPrice per generated clip
3 seconds$1.2600
4 seconds$1.6800
5 seconds$2.1000
6 seconds$2.5200
7 seconds$2.9400
8 seconds$3.3600
9 seconds$3.7800
10 seconds$4.2000
11 seconds$4.6200
12 seconds$5.0400
13 seconds$5.4600
14 seconds$5.8800
15 seconds$6.3000

Billing follows params.duration * 0.42: multiply generated seconds by the $0.4200 per-second rate.

When to Use

✅ Good fit❌ Consider alternatives
The model's named workflow matches the source material and intended outputA different input modality or model route is required
A managed asynchronous result is suitable for the production pipelineA synchronous, interactive editor is essential
The documented controls cover the required duration, framing, or formatThe project needs controls outside this endpoint's schema
Creative iteration benefits from a repeatable request structureExact deterministic pixels, frames, geometry, or samples are mandatory
A finished downloadable media asset is the desired deliverableEditable source layers or a native project file are required

Prompt Guide

For image-conditioned generation, state the intended result first, then add the subject or source treatment, progression, style, and delivery constraints. Keep one creative variable per phrase, use the documented field names for controls, and change one setting at a time when comparing results.

{
  "aspect_ratio": "16:9",
  "duration": 8,
  "image": "https://static.sandbase.ai/examples/mirrored/d5f4a0f0f119-NaslJIC7F2WodS6DFZRRJ.png",
  "images": [
    [
      "https://example.com/reference.png"
    ]
  ],
  "prompt": "@Element1 and @Element2 enters the scene from two sides. Elephant starts to play with the ball"
}

Technical Specs

SpecValue
Model IDkwaivgi/kling-video/o3/4k/reference-to-video
Inputsaspect_ratio, duration, end_image, generate_audio, image, images, multi_prompt, prompt, shot_type
Required inputsprompt, image
Output fieldscontent_type, url
ExecutionAsync (submit, then poll for result)
Duration3 / 4 / 5 / 6 / 7 / 8 / 9 / 10 / 11 / 12 / 13 / 14 / 15
Aspect Ratio16:9 / 9:16 / 1:1

Related Models

Start building

Send your first request

OpenAI-compatible endpoint with unified authentication and usage tracking.

Production API
Unified Run API endpoint
https://api.sandbase.ai/v1/run
Model ID
kwaivgi/kling-video/o3/4k/reference-to-video
# The quoted heredoc keeps Unicode and shell metacharacters unchanged.
result=$(curl --fail-with-body --silent \
  -X POST "https://api.sandbase.ai/v1/run" \
  -H "Authorization: Bearer $SANDBASE_API_KEY" \
  -H "Content-Type: application/json" \
  --data-binary @- <<'SANDBASE_JSON'
{
  "model": "kwaivgi/kling-video/o3/4k/reference-to-video",
  "image": "https://static.sandbase.ai/examples/mirrored/d5f4a0f0f119-NaslJIC7F2WodS6DFZRRJ.png",
  "prompt": "@Element1 and @Element2 enters the scene from two sides. Elephant starts to play with the ball",
  "duration": 8,
  "end_image": "https://static.sandbase.ai/examples/mirrored/b5db8ef4ab4d-BwHi22qoQnqaTNMMhe533.png",
  "shot_type": "customize",
  "generate_audio": false
}
SANDBASE_JSON
)
run_id=$(printf '%s' "$result" | jq -r .id)
for attempt in $(seq 1 120); do
  status=$(printf '%s' "$result" | jq -r .status)
  case "$status" in completed|failed|timeout) break ;; esac
  sleep 2
  result=$(curl --fail-with-body --silent \
    -H "Authorization: Bearer $SANDBASE_API_KEY" \
    "https://api.sandbase.ai/v1/run/$run_id")
done
status=$(printf '%s' "$result" | jq -r .status)
[ "$status" = completed ] || { echo "Generation ended: $status" >&2; exit 1; }
printf '%s\n' "$result"

Choose your model

Compare models

All language models
ModelContextInput / 1MOutput / 1MReleased
KwaiVGI———Apr 23, 2026
KwaiVGI———Jul 6, 2026
KwaiVGI———Jul 6, 2026
KwaiVGI———Jul 3, 2026
KwaiVGI———Jul 3, 2026
KwaiVGI———Jul 3, 2026