Kling 3.0 Turbo is an image-to-video model from Kling AI, now live on EmpirioLabs. Give it one source image and a short prompt describing the motion, and it animates the image into a smooth video with synchronized native audio. It renders at 720p or 1080p for 3 to 15 seconds, and it is built for speed, so clips come back quickly.
It is image-to-video only: an image is required on every request, and a text prompt is optional but recommended to direct the motion. Native audio is generated automatically, so the output already has sound.
Pricing
Kling 3.0 Turbo is pay-as-you-go, billed per second of video generated, with 1080p priced higher per second than 720p. There is no subscription and no minimum, and a request that fails before any video is produced is not billed. The live per-second rates are on the Kling 3.0 Turbo model page and the pricing page.
Quickstart
The API is OpenAI-compatible. Submit a generation, then poll the returned job until it is ready. Set your key as EMPIRIOLABS_API_KEY and call the videos endpoint:
curl https://api.empiriolabs.ai/v1/videos/generations \
-H "Authorization: Bearer $EMPIRIOLABS_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "kling-3-0-turbo",
"image": "https://your-host.com/source.jpg",
"prompt": "cinematic slow push-in, gentle wind, drifting clouds",
"resolution": "720p",
"duration": 5
}'The response returns a job_id and a poll_url. Poll it until the status is completed:
curl https://api.empiriolabs.ai/v1/jobs/JOB_ID \
-H "Authorization: Bearer $EMPIRIOLABS_API_KEY"When it completes, the response includes a signed video URL you can download. The source image can be passed as a public URL or as base64, in JPG or PNG.
Parameters
- image (required): the source image to animate.
- prompt: describes the motion to apply.
- resolution: 720p or 1080p.
- duration: video length in seconds, 3 to 15.
Good to know
- Image-to-video only, so an image is always required. Use a real photo or rendered frame; logos or graphics with transparent backgrounds may be rejected.
- Native audio is always on, so the clip arrives with sound.
- Generation is asynchronous: submit, then poll the job. Typical clips finish in well under a minute.
- You only pay for video that is produced; failed requests are not billed.
Multi-shot prompts
Kling 3.0 Turbo can render up to 6 sequential shots in a single video. Format the prompt as shot 1, <seconds>, <description>; shot 2, <seconds>, <description>; where each shot lasts at least 1 second and the shot durations add up to the total duration. For example, a 6 second video: shot 1, 3, a lone astronaut walks across a red desert at dawn; shot 2, 3, the camera tilts up to reveal two pale moons;
Try it in the playground, read the API docs, or open the model page.
