kling-v3 is the standard video generation model in Kuaishou Kling 3.0, suitable for creating short films from text concepts or still images. It supports flexible whole-second durations, first/last-frame guidance, synchronized audio, and multiple aspect ratios, making it suitable for both advertising shots and social content, as well as previewing motion in scenes. When you need to directly edit existing videos or combine multiple reference materials, choose a different Omni model.
Submit requests to the public API at api.acedata.cloud using the documented parameters, then use the results in your application.
Input parameters and result formats vary by service. Use the public API for this model and follow its guide for generation, task retrieval and editing operations.
Clarify capacity, inputs and outputs, and invocation methods before selecting a model.
The above are the invocation specifications for kling-v3 on this platform. Native creation capabilities and available parameters for each mode should be understood separately.
Learn what kling-v3 can bring to your work.
Text-to-video is suitable for designing scenes from scratch: use prompts to specify the subject, action, environment, lighting, and camera intent, then choose the duration and aspect ratio. Relevance strength and negative prompts can help express creative constraints, making this suitable for iteratively refining the same shot rather than cramming an entire complex storyline into one short clip.
Image-to-video uses a first-frame image to establish the visual starting point, and an end frame can be added to specify the final image. For product showcases, character actions, or scene transitions, this approach can turn static designs into dynamic assets. Prompts should focus on how motion unfolds between the two images and avoid contradicting the subjects and composition in the images.
When a sound-enabled short clip is needed, synchronized audio generation can be enabled; when more explicit camera movement is needed, camera control can be used in std or pro mode. 4k mode is suitable for tasks that prioritize output clarity, but camera_control cannot be added at the same time, so image-quality settings and camera control should be weighed before creation.
Start with specific tasks to find where the model can be effective.
Input a static product image, describe the display action, background atmosphere, and camera changes, and generate product shots for subsequent editing. If a clear final composition already exists, an end frame can also be provided. It is recommended to clearly specify the featured subject and core action, complete individual shots first, and then assemble them into a complete advertisement.
Use the event theme, setting, and subject action as text input, select a 9:16 aspect ratio and an appropriate whole-second duration, and create vertical short-video assets. Enable generate_audio when sound is needed; when brand posters or visual designs already exist, use image-to-video to bring the design into dynamic expression.
Turn storyboard descriptions or keyframes into short clips for discussing action pacing, scene atmosphere, and camera direction. Keep subject and scene descriptions consistent, adjust only one key factor per round, and make it easier to compare different options. The deliverable is video footage suitable for further editing, rather than completing all post-production in one pass.
Choose based on task complexity, input materials, and expected results.
If the task is “generate a shot from a piece of text” or “make an image move,” kling-v3 offers a more direct workflow, along with first-and-last-frame and camera movement options. If you need to use multiple assets as independent references, or add or remove elements from existing videos and modify their style, choose kling-v3-omni or kling-o1; the standard version's first-and-last-frame guidance is not equivalent to video editing.
Compared with most older models that only accept 5 or 10 seconds, kling-v3 lets you set shots from 3–15 seconds in whole-second increments, making it easier to match the pacing of short videos. It supports synchronized audio, while kling-v2-6 limits audio to pro. Existing older-version workflows can be retained; prioritize this model when you need flexible duration, 4K, or a combination of camera movement and audio.
From a small-scale task to production integration.
Define the goal, required inputs, and output requirements, using real business examples as a starting point.
Open the trial page, confirm the parameters supported by this endpoint, then submit a small-scale task to review the results.
Keep the complete model ID, use the request format specified in the documentation, and confirm billing rules on the Pricing page.
Before formal use, understand the output quality and capability scope.
Answers to common questions about using kling-v3.
Submit a request to POST /kling/videos and explicitly set model to kling-v3. For text creation, use action=text2video and fill in prompt; for image creation, use image2video and provide start_image_url, then choose duration, aspect ratio, and audio as needed.
Integer durations from 3–15 seconds are supported, along with three aspect ratios: 16:9, 9:16, and 1:1. You can compose shots for ads, vertical short videos, or square content; the duration should match the subject's actions, and too many continuous events should not be arranged within shorter clips.
Select image2video and provide the starting frame start_image_url first, then add end_image_url when you need to specify the ending image. The end frame cannot be submitted without the start frame. The prompt can describe how the subject transitions from the initial state to the final state, keeping the visual guidance consistent with the action requirements.
4k mode supports generate_audio, but does not support camera_control. Explicitly enable audio generation when you need sound; use std or pro when you need parameterized camera movement. Audio is disabled by default, so selecting 4k alone will not automatically produce a video with sound.
You can set async=true to first obtain task_id, then query task progress and results; you can also configure callback_url to receive completion notifications. Completed results include information such as video_url, video_id, and state. Retrieve the video only after confirming success, rather than treating successful task submission as completed video generation.
Model information · Updated: 2026-10-01. See the API and pricing sections for invocation parameters and billing rules.
Choose a development approach or application suited to this model and follow its setup guide.
↗Python SDK Integration Guide
SDK installation, authentication and examples↗TypeScript SDK Integration Guide
SDK installation, authentication and examples
↗Pricing
$0.0799–$0.40· per second
You pay$13,584for 142,800 credits · $0.0951 / credits46% OFF
You're billed in credits. Pick a top-up tier to see its per-credit rate — the larger the top-up, the cheaper each credit.
This is only a basic call example. See the full docs for more parameters and advanced usage.
View Docs