Choose a category
3 models available
8 models available
3 models available
5 models available
6 models available
3 models available
8 models available
Kling 3.0 from a prompt: 3–15 seconds, up to 4K, with optional native audio.
Describe the subject, action and camera move. Put dialogue in quotation marks when audio is on. Up to 2,500 characters.
Choose 3–15 seconds. Priced by the second.
Standard is the cheapest tier. 4K costs several times more per second.
Adds dialogue, sound effects and ambience. Costs more per second at Standard and Pro.
Write a scene and Kling 3.0 films it: 3 to 15 seconds in Standard, Pro or 4K quality, in 16:9, 9:16 or 1:1. Switch on native audio and the model generates dialogue, sound effects and ambience together with the picture, so a spoken line in your prompt is actually spoken in the clip.
Put the exact words a character should say in quotation marks and turn Generate Audio on.
Sound, a 4K tier and second-by-second duration.
With Generate Audio on, Kling 3.0 produces the soundtrack in the same pass as the video: voices, footsteps, weather, room tone. Quote a line of dialogue and name who says it, and the character's mouth movement is generated to match.
Standard is the economical tier for drafts and social clips. Pro renders a sharper, higher-resolution picture for finished work. 4K is the top tier for footage that will be shown large or cropped in an edit. The tier changes the per-second cost, shown before you submit.
Set the exact number of seconds rather than choosing between two fixed lengths. A 3-second insert, a 7-second product beat and a 15-second scene all come from the same slider.
Both are on this site. Turbo is the silent sibling at 720p or 1080p. This page is the full Kling 3.0 model: choose it when you need generated audio, the 4K tier or, on the image page, a fixed last frame.
Change one part of the input at a time so you can judge what helped.
Scene, performance, camera, sound.
Start with place, light and who is in frame: “A barista in a sunlit cafe pours steamed milk into a cup.” Kling 3.0 follows physical actions closely, so describe what hands, faces and objects actually do.
Add one camera instruction — slow push-in, handheld follow, static wide — and, if audio is on, the line to be spoken in quotation marks. Prompts can run to 2,500 characters, but one clear shot beats a paragraph of alternatives.
Pick Standard, Pro or 4K, the number of seconds, the aspect ratio and whether to generate audio. The credit cost updates as you change them. Sign in, submit, and find the finished clip in your history.
Change one part of the input at a time so you can judge what helped.
An original Kling 3.0 text-to-video result generated on Flow Veo 3 at Pro quality with Generate Audio on. Play it to hear the line from the prompt. No images were supplied.
Prompt: A barista in a sunlit cafe pours steamed milk into a cup, forming latte art. Slow push-in on the cup as steam rises. She looks up, smiles and says "Your flat white is ready." Ambient cafe chatter and the hiss of the espresso machine.
Answers about this workflow, inputs and output settings.
Kling 3.0 is a video generation model from Kuaishou's Kling AI. This page runs its text-to-video workflow on Flow Veo 3, an independent service that is not affiliated with Kuaishou or Kling AI.
Yes. Turn on Generate Audio and the clip comes with dialogue, sound effects and ambience generated alongside the picture. With the switch off the video is silent and, at Standard and Pro quality, costs less per second.
Yes. Choose 4K in the Quality menu. It is the most expensive tier per second, so it is worth confirming the shot at Standard first and re-running the prompt in 4K once you are happy with it.
Any whole number of seconds from 3 to 15. Pricing is per second, so a 15-second clip costs five times a 3-second one at the same quality.
Turbo is silent and offers 720p or 1080p. Kling 3.0 adds native audio and a 4K tier, and its image-to-video page accepts a last frame as well as a first frame. Both are available from the model menu.
Not on this page yet. Each generation is a single shot of 3 to 15 seconds. To build a sequence, generate the shots separately and join them in your editor.
Credits are held when you submit and released automatically if the generation fails. Finished videos are only available for 7 days, so download the ones you want to keep.
Choose the workflow that matches your source material.