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,@Video1and@Audio1in 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:
| Resolution | Price 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 $0.282 at 1080p.
Parameters
| Parameter | Description |
|---|---|
| prompt | Required. Refer to images as @Image1, @Image2 …, videos as @Video1 … and audios as @Audio1 … |
| images | Reference images, up to 10. |
| videos | Reference videos, up to 5. Together they are used for up to 30 seconds; longer clips are trimmed. |
| audios | Reference audios, up to 3, 5 to 30 seconds each. |
| keep_original_sound | Keep the reference video’s own sound instead of generating new audio (default false). Requires videos. |
| resolution | 720p (default) or 1080p. |
| duration | 3 to 30 seconds, default 5. |
| aspect_ratio | auto (default, follows the references), 16:9, 9:16, 1:1 or 21:9. |
Prompting Guide
- Say what each reference is for: “the character from @Image1 walks through the street from @Image2”.
- To reuse a camera move, say so: “follow the camera movement of @Video1”.
- Use clear, well-lit reference images; the subject should be easy to recognize.
Notes
- Reference audios guide a speaker’s voice; they do not reproduce the original recording, music, or timing.
- Set
keep_original_soundtofalseto generate new audio using the reference voice. Whentrue, the reference video’s original audio is retained.
Authentication
For authentication details, please refer to the Authentication Guide.
API Endpoints
Submit Task & Query Result
set -euo pipefail
export WAVESPEED_API_KEY="your-api-key"
REQUEST_BODY=$(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
doneRetry temporary errors only while polling, never the submit request. Errors, retries and timeouts
Parameters
Task Submission Parameters
Request Parameters
| Parameter | Type | Required | Default | Range | Description |
|---|---|---|---|---|---|
| prompt | string | Yes | - | 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, ... | |
| images | array<string> | No | - | 0 ~ 10 items | Reference images (up to 10). |
| videos | array<string> | No | - | 0 ~ 5 items | Reference videos (up to 5). They guide motion, camera and pacing; together they are used for up to 30 seconds. |
| audios | array<string> | No | - | 0 ~ 3 items | Reference audios (up to 3), 5-30 seconds each. They guide the voice and sound of the video. |
| keep_original_sound | boolean | No | false | - | Keep the reference video's own sound instead of generating new audio. Requires videos. |
| resolution | string | No | 720p | 720p, 1080p | Output resolution. |
| duration | integer | No | 5 | 3, 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, 30 | The duration of the generated video in seconds (3-30). |
| aspect_ratio | string | No | auto | auto, 16:9, 9:16, 1:1, 21:9 | The 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
| Parameter | Type | Description |
|---|---|---|
| code | integer | HTTP status code (e.g., 200 for success) |
| message | string | Status message (e.g., “success”) |
| data.id | string | Unique identifier for the prediction, Task Id |
| data.model | string | Model ID used for the prediction |
| data.outputs | array | Output values, usually URL strings; some models return text strings or structured result objects (empty when status is not completed) |
| data.urls | object | Object containing related API endpoints |
| data.status | string | Task status. completed is successful; failed, cancelled, timeout, and deleted are failure terminal statuses. |
| data.created_at | string | ISO timestamp of when the request was created (e.g., “2023-04-01T12:34:56.789Z”) |
| data.error | string | Error message (empty if no error occurred) |
| data.timings | object | Object containing timing details |
| data.timings.inference | integer | Inference time in milliseconds |
Result Request Parameters
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
| id | string | Yes | - | Task ID |
Result Response Parameters
| Parameter | Type | Description |
|---|---|---|
| code | integer | HTTP status code (e.g., 200 for success) |
| message | string | Status message (e.g., “success”) |
| data | object | The prediction data object containing all details |
| data.id | string | Unique identifier for the prediction |
| data.model | string | Model ID used for the prediction |
| data.outputs | array<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.urls | object | Object containing related API endpoints |
| data.status | string | Status: completed is successful; failed, cancelled, timeout, and deleted are failure terminal statuses |
| data.created_at | string | ISO timestamp of when the request was created |
| data.error | string | Error message (empty if no error occurred) |
| data.timings | object | Object containing timing details |
| data.timings.inference | integer | Inference time in milliseconds |