Introduction
Kling 4.0 Flash video-edit starts from an existing clip and a required instruction about what to change. Supply a source of 3–20 seconds and, if useful, up to 10 reference images. The 720p result keeps the source’s scene, subject, motion, and camera move while applying the described edit.
Output duration follows the source, and sources longer than 20 seconds are cut at that point. In our tests, the observatory wardrobe edit kept a 5.04-second source at 5.04 seconds. Source sound is retained by default; set keep_original_audio to false for newly generated synchronized audio.
Key features & highlights
- Build a visual revision around an existing performance and camera move.
- Add up to 10 reference images when the edit calls for a particular visual reference.
- Keep source timing in the output, with source material capped at 20 seconds.
- Retain the original soundtrack by default or generate fresh synchronized audio.
- Receive 720p output in auto framing or one of four supported aspect ratios.
Request parameters
| Parameter | Required | Description |
|---|---|---|
prompt | Yes | Required edit instructions with no preset text; describe the change and the scene, subject, motion, and camera move to retain. The instruction itself adds no charge. at least 1 characters |
video_url | Yes | Required URL for one source clip of 3–20 seconds, with no preset source; longer inputs are cut at 20 seconds. at least 1 characters |
reference_image_urls | No | Optionally add up to 10 image URLs to guide the edit, identifying them as @Image1, @Image2, and so on in upload order. Image references add no separate charge. at least 1 items · at most 10 items |
aspect_ratio | No | Use auto to follow the source video, or set 16:9, 9:16, 1:1, or 21:9 for the output shape. This setting does not change the edit charge. |
keep_original_audio | No | Defaults to true to keep source sound; false generates a new synchronized soundtrack. Switching this value adds no separate charge. Default: |
Pricing
$0.151 / second
Editing is charged per source second rounded up to the next whole second, with a maximum of 20 billed seconds. A 5.04-second source counts as 6 billed seconds, even when the output stays 5.04 seconds. The prompt, reference images, aspect ratio, and audio choice add no separate charge. Kling moderation refusals are not billed.
Use cases
- Try an emerald evening gown in an observatory clip while keeping the action.
- Review different coat colors across versions of a short social scene.
- Prepare visual draft variations of a recorded movement with the source soundtrack retained.
Prompt guide & pro tips
- Start with what to preserve, such as “retain the scene and camera movement,” to define the edit’s fixed elements.
- Name a focused change, such as “make the coat charcoal gray,” before requesting several variations.
- Assign a role to each added image, such as “use the jacket shown in @Image1,” so the reference relates to the change.
- Describe material and color together, such as “an emerald velvet gown,” when the wardrobe is the subject of the edit.
- For a fresh soundtrack, turn off
keep_original_audioand include a cue such as “Sound: a low motor hum and evening wind.”
API quick start
# Generate the idempotency key once; reuse it when retrying after a lost response.
export SPICY_IDEMPOTENCY_KEY="${SPICY_IDEMPOTENCY_KEY:-$(uuidgen)}"
curl -X POST https://api.spicyapi.ai/api/v1/jobs/createTask \
-H "Authorization: Bearer $SPICY_API_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: $SPICY_IDEMPOTENCY_KEY" \
-d '{
"model": "kling/4.0-flash/video-edit",
"input": {
"aspect_ratio": "auto",
"prompt": "A slow aerial shot drifting over a misty pine forest at sunrise",
"video_url": "https://example.com/reference.mp4"
}
}'
# Copy data.taskId from the response into TASK_ID, then poll the status.
TASK_ID="TASK_ID_FROM_RESPONSE"
curl "https://api.spicyapi.ai/api/v1/jobs/recordInfo?taskId=$TASK_ID" \
-H "Authorization: Bearer $SPICY_API_KEY"
See the API reference for authentication, the complete schema, polling, and error handling.
