kling-v2-6 is Kuaishou Kling's V2.6 video model, suitable for turning text concepts or static images into short videos. Its key value is that pro mode provides both first and last frame control and synchronized audio, making it useful for product showcases, atmospheric clips, and narrative shots; in talking-photo workflows, it can also combine portrait photos with existing audio to create lip-synced videos.
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.
The above are this platform's V2.6 invocation specifications. Photo lip-sync is a separate workflow and is not the same as synchronized audio during video generation.
V2.6's pro mode supports enabling audio while generating video, making it suitable for short-form creations that need both visuals and sound. Prompts can be organized around the subject, actions, and scene, then audio can be enabled with generate_audio; when only visual assets are needed, keep it disabled for easier later dubbing and editing.
Image-to-video can start from a specified first frame, and pro mode can also add a last frame to set clear opening and closing visuals for a clip. It is suitable for projects with existing product images, character images, or storyboards, allowing creation to develop around the selected composition; first and last frames constrain visual boundaries rather than providing frame-by-frame editing of every intermediate frame.
The talking-photo workflow accepts a portrait image and existing audio, first animates the photo, then completes audio lip-sync, delivering the final video in a single submission. You can use a prompt to describe movements and expressions during the animation stage, and explicitly select kling-v2-6; this workflow is suited to tasks with existing recordings rather than directly synthesizing speech from text.
Provide a main product image and a description of the display action to create short videos suitable for landscape, portrait, or square publishing. If you have already designed the closing shot, use pro to add an end frame so the product showcase lands on the intended composition; the delivered video link can enter the editing workflow, where you can further add brand text, subtitles, and a complete marketing arrangement.
Starting from descriptions of scenes, character actions, and events, use text-to-video to create short shots, and enable synchronized audio in pro mode. This is suitable for first validating a story segment or soundscape concept before deciding whether to move into formal editing; organizing the task into clear individual clips makes checking easier than cramming an entire long-form script into a single generation.
Prepare a clear front-facing photo of one person, along with a recording that matches the target video length, and submit them through the talking photo entry point. The output includes the final lip-synced video, and you can also obtain the video link from the photo animation stage, making it easy to separately check character movement and lip-sync results. It is suitable for character introductions, brief explanations, and talking-head samples.
If you need first and last frames, the pro modes of both can be considered; if you want audio at the same time as generation, V2.6 pro better meets the task requirements, while V2.5 Turbo does not provide this feature. For clips that need neither sound nor an end frame, start with V2.6 std; choose pro when sound or a clearly defined closing shot is needed, rather than judging solely by the version name.
V2.6 is suitable for fixed short durations, text- or first-frame-driven generation, and audio creation in pro mode. If the task requires integer durations of 3–15 seconds, native 4K mode, or dedicated camera movement controls, consider the corresponding V3 capabilities; if you need to combine multiple reference images, reference videos, or directly edit existing videos, choose O1 or V3 Omni rather than applying these capabilities to V2.6.
text2video starts from a prompt; image2video provides start_image_url. If you need to specify the ending composition, you can also provide end_image_url under pro.
Specify model=kling-v2-6 for /kling/videos, and choose 5 or 10 seconds. When synchronized audio is needed, use mode=pro and generate_audio=true; for silent tasks, audio can remain disabled.
Save the async task ID or use callback_url to retrieve results; check the subject, action, and ending. For tasks with audio enabled, also listen to the dialogue, sound effects, and timing; when not enabled, proceed to editing as silent footage.
Keep the appearance of the coffee machine in the reference image, generate a shot of coffee being poured into a cup, with the sound of the machine operating and coffee flowing, and keep the image stable.
When synchronized audio is needed, choose pro and generate_audio=true, and listen to the mechanical sound; std does not inherit pro's audio or end-frame capabilities.
If you only create clips without synchronized audio and do not need to specify an end frame, you can choose std. If you need synchronized audio or start/end frame control, use pro. Both modes support 5-second or 10-second videos; these pro features must be enabled explicitly and are not all automatically turned on simply by selecting this mode.
In the /kling/videos request, explicitly set model to kling-v2-6, mode to pro, and generate_audio to true. This switch is off by default. Synchronized audio and submitting existing audio for photo lip-sync are two different approaches; choose based on whether you already have a recording.
You cannot use an end frame alone as input for this image-to-video workflow. Use action=image2video, provide start_image_url, and add end_image_url as needed in pro mode. If you only have one image, use it as the start frame and describe the desired action in the prompt.
Submit image_url and audio_url to /kling/talking-photo, explicitly specify model=kling-v2-6, and choose 5 seconds or 10 seconds. The prompt can describe the movements and expressions during the photo animation stage. The final video_url is the lip-synced video, while source_video_url is the intermediate photo animation video.
You can set async=true, save the returned task_id first, then retrieve the status and result through task queries; you can also configure callback_url to receive completion notifications. After obtaining video_url, download it or proceed to the editing workflow. Both generation endpoints should explicitly specify kling-v2-6 to avoid using other default models.
Choose a development approach or application suited to this model and follow its setup guide.
Python SDK Integration Guide
SDK installation, authentication and examplesTypeScript SDK Integration Guide
SDK installation, authentication and examples
