Kwaivgi Kling V4.0 Reference To Video API Documentation

Kwaivgi Kling V4.0 Reference To Video API Documentation

Playground

Try it on WaveSpeedAI!

Kling V4.0 Reference-to-Video builds a new clip of 3 to 30 seconds from up to 10 reference images, 5 reference videos and 3 reference audios, following the subjects, motion and camera you point to in the prompt, with 720p and 1080p output and synchronized stereo audio. Ready-to-use REST inference API, best performance, no coldstarts, affordable pricing.

Features

Kling V4.0 Reference-to-Video creates a new clip of 3 to 30 seconds from up to 10 reference images, up to 5 reference videos and up to 3 reference audios. Point to them in the prompt to keep a subject consistent, borrow a camera move or follow a performance. Output is 720p or 1080p with synchronized stereo audio.


Why Choose This?

  • Mix images, videos and audio Combine up to 10 images, 5 videos and 3 audios (15 files at most) in one request.
  • Prompt-level control Refer to inputs as @Image1, @Video1 and @Audio1 in the order you provide them.
  • Up to 30 seconds Output length is yours to choose, independent of the reference video.
  • Native audio New synchronized stereo audio, or keep the reference video’s own sound.

Pricing

Without a reference video:

ResolutionPrice per second
720p$0.157
1080p$0.235

With reference videos, the price is per second counted as the output length plus the longer of the output length and the total length of the reference videos (in whole seconds, up to 30). Reference audios are not charged. Each counted second costs $0.094 at 720p and $0.141 at 1080p. When the output is as long as the reference videos this is $0.188 per output second at 720p and $0.282 at 1080p.


Parameters

ParameterDescription
promptRequired. Refer to images as @Image1, @Image2 …, videos as @Video1 … and audios as @Audio1 …
imagesReference images, up to 10.
videosReference videos, up to 5. Together they are used for up to 30 seconds; longer clips are trimmed.
audiosReference audios, up to 3, 5 to 30 seconds each.
keep_original_soundKeep the reference video’s own sound instead of generating new audio (default false). Requires videos.
resolution720p (default) or 1080p.
duration3 to 30 seconds, default 5.
aspect_ratioauto (default, follows the references), 16:9, 9:16, 1:1 or 21:9.

Prompting Guide

  • Say what each reference is for: “the character from @Image1 walks through the street from @Image2”.
  • To reuse a camera move, say so: “follow the camera movement of @Video1”.
  • Use clear, well-lit reference images; the subject should be easy to recognize.

Notes

  • Reference audios guide a speaker’s voice; they do not reproduce the original recording, music, or timing.
  • Set keep_original_sound to false to generate new audio using the reference voice. When true, the reference video’s original audio is retained.

Authentication

For authentication details, please refer to the Authentication Guide.

API Endpoints

Submit Task & Query Result

set -euo pipefail

export WAVESPEED_API_KEY="your-api-key"

REQUEST_BODY=$(cat <<'JSON'
{
  "prompt": "A cinematic ocean wave at sunrise, highly detailed",
  "keep_original_sound": false,
  "resolution": "720p",
  "duration": 5,
  "aspect_ratio": "auto"
}
JSON
)

# 1. Submit the prediction once. Never retry this request automatically:
# a retry after an unclear failure can create a second, billed task.
CHECK='The task may already exist. Check History or the predictions list before submitting again.'
if ! SUBMIT_RESPONSE=$(curl --silent --show-error --fail-with-body \
  --connect-timeout 10 --max-time 60 \
  -X POST "https://api.wavespeed.ai/api/v3/kwaivgi/kling-v4.0/reference-to-video" \
  -H "Authorization: Bearer ${WAVESPEED_API_KEY}" \
  -H "Content-Type: application/json" \
  -d "${REQUEST_BODY}"); then
  printf 'Submit failed. %s\n' "${CHECK}" >&2
  exit 1
fi

TASK=$(printf '%s' "${SUBMIT_RESPONSE}" | jq 'if type == "object" and has("data") then .data else . end')
PREDICTION_ID=$(printf '%s' "${TASK}" | jq -r '.id // empty')
if [ -z "${PREDICTION_ID}" ]; then
  printf 'Submission response did not contain a prediction id. %s\n' "${CHECK}" >&2
  exit 1
fi
RESULT_URL="https://api.wavespeed.ai/api/v3/predictions/${PREDICTION_ID}/result"

# 2. Poll until the prediction reaches a final status.
BODY_FILE=$(mktemp)
trap 'rm -f "${BODY_FILE}"' EXIT
while true; do
  # Transient errors (429, 5xx) are retried, honoring Retry-After.
  HTTP_CODE=$(curl --silent --show-error --fail-with-body \
    --connect-timeout 10 --max-time 60 --retry 5 --retry-delay 2 \
    --output "${BODY_FILE}" --write-out '%{http_code}' \
    "${RESULT_URL}" \
    -H "Authorization: Bearer ${WAVESPEED_API_KEY}" || true)
  case "${HTTP_CODE}" in
    2??) ;;
    000|429|5??)  # dropped connection or temporary error: keep polling
      printf 'Task %s: HTTP %s, polling again in 10s\n' "${PREDICTION_ID}" "${HTTP_CODE}" >&2
      sleep 10; continue ;;
    *) printf 'Task %s: HTTP %s\n' "${PREDICTION_ID}" "${HTTP_CODE}" >&2
      cat "${BODY_FILE}" >&2; exit 1 ;;
  esac
  RESULT=$(jq 'if type == "object" and has("data") then .data else . end' "${BODY_FILE}")
  STATUS=$(printf '%s' "${RESULT}" | jq -r '.status // empty')

  case "${STATUS}" in
    completed) printf '%s\n' "${RESULT}" | jq '.outputs'; break ;;
    failed|cancelled|timeout|deleted)
      printf 'Task %s ended with status %s\n' "${PREDICTION_ID}" "${STATUS}" >&2
      printf '%s\n' "${RESULT}" | jq . >&2; exit 1 ;;
    *) sleep 2 ;;
  esac
done

Retry temporary errors only while polling, never the submit request. Errors, retries and timeouts

Parameters

Task Submission Parameters

Request Parameters

ParameterTypeRequiredDefaultRangeDescription
promptstringYes-The positive prompt for the generation. Refer to reference images as @Image1, @Image2, ... (in array order), to reference videos as @Video1, @Video2, ... and to reference audios as @Audio1, @Audio2, ...
imagesarray<string>No-0 ~ 10 itemsReference images (up to 10).
videosarray<string>No-0 ~ 5 itemsReference videos (up to 5). They guide motion, camera and pacing; together they are used for up to 30 seconds.
audiosarray<string>No-0 ~ 3 itemsReference audios (up to 3), 5-30 seconds each. They guide the voice and sound of the video.
keep_original_soundbooleanNofalse-Keep the reference video's own sound instead of generating new audio. Requires videos.
resolutionstringNo720p720p, 1080pOutput resolution.
durationintegerNo53, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30The duration of the generated video in seconds (3-30).
aspect_ratiostringNoautoauto, 16:9, 9:16, 1:1, 21:9The aspect ratio of the generated video; auto follows the reference images or video.

Every model also accepts the global request options enable_sync_mode and enable_base64_output; see their pages for details.

Response Parameters

ParameterTypeDescription
codeintegerHTTP status code (e.g., 200 for success)
messagestringStatus message (e.g., “success”)
data.idstringUnique identifier for the prediction, Task Id
data.modelstringModel ID used for the prediction
data.outputsarrayOutput values, usually URL strings; some models return text strings or structured result objects (empty when status is not completed)
data.urlsobjectObject containing related API endpoints
data.statusstringTask status. completed is successful; failed, cancelled, timeout, and deleted are failure terminal statuses.
data.created_atstringISO timestamp of when the request was created (e.g., “2023-04-01T12:34:56.789Z”)
data.errorstringError message (empty if no error occurred)
data.timingsobjectObject containing timing details
data.timings.inferenceintegerInference time in milliseconds

Result Request Parameters

ParameterTypeRequiredDefaultDescription
idstringYes-Task ID

Result Response Parameters

ParameterTypeDescription
codeintegerHTTP status code (e.g., 200 for success)
messagestringStatus message (e.g., “success”)
dataobjectThe prediction data object containing all details
data.idstringUnique identifier for the prediction
data.modelstringModel ID used for the prediction
data.outputsarray<string | object>Array of generated outputs (empty when status is not completed). Items are usually URL strings, but may be text strings or structured result objects, depending on the model.
data.urlsobjectObject containing related API endpoints
data.statusstringStatus: completed is successful; failed, cancelled, timeout, and deleted are failure terminal statuses
data.created_atstringISO timestamp of when the request was created
data.errorstringError message (empty if no error occurred)
data.timingsobjectObject containing timing details
data.timings.inferenceintegerInference time in milliseconds
© 2026 WaveSpeedAI. All rights reserved.