KLING AI says it can turn text, images, and references into multimodal creative content in one studio. The homepage names Kling 4.0, Kling Image 3.0, Canvas, and an MCP / CLI batch path beside image, video, and motion-control menus. The VIDEO 3.0 guide adds a 3 to 15 second duration, two multi-shot switches, native audio, and credit examples. A text-to-video FAQ is the only public page with a dollar figure for paid plans.
Official website · kling.ai
Choose KLING AI when the clip has a control the pages actually name: a prompt, a start frame, element references, or one person’s motion reference. The guide is specific about duration and credits. The motion page is specific about a single primary subject. Skip it as a priced membership comparison. Named plan cards were not verified on the pages reviewed here. A free 720p tier and a paid floor appear only as FAQ sentences. Homepage films, view counts, and the 60M+ user line are the site’s own credits, not a measurement from this directory.
The homepage says the product will turn text, images, and references into multimodal creative content in one studio. Menus name image generation, video generation, motion control, and an agent, then text to video, image to video, an AI video editor, text to image, image to image, an AI photo editor, and an AI avatar. Kling 4.0 opens with “You call the shots” and the sample line “A pianist plays on a ship amid towering waves,” which is a prompt on the page, not a rendered result. Kling Image 3.0 covers stills: cinematic language, native 4K frames, and series generation that keeps a style across a set. On Canvas you brainstorm with an agent, write scripts, and generate images, videos, and storyboards, then refine and edit your videos through later passes. Skills named there are storyboards, posters, and e-commerce assets, run in parallel, with nodes you can rearrange. The MCP / CLI block says you can generate multiple versions at once to test styles and batch-produce videos. “Multi-Task Production” and “Smart Analysis & Generate” without command syntax. The quickstart index adds guide names rather than a shared form: VIDEO 3.0, IMAGE 3.0, both Omni variants, Element Library 3.0, VIDEO O1, IMAGE O1, Avatar 2.0, and VIDEO 2.6. Essential labels are Motion Control, Text-to-Video, Image-to-Video, Elements, and Extend with Prompts. Advanced labels are Standard Mode and Professional Mode, Camera Movement, Start and End Frames, Motion Brush, and Lip sync. Downloads listed there are iOS, Android, macOS M Chip, macOS Intel Chip, and Windows. A footer cluster reads API Platform About Us Contact Us support@kling.ai. The developer pricing capture was too short to quote a unit price.
Official documentationThe VIDEO 3.0 guide builds on VIDEO O1 and VIDEO 2.6. The new model generates up to 15 seconds of continuous video, with a flexible duration ranging from 3 to 15 seconds. A table marks Multi-Shot, start frame plus element reference, multi-character coreference of three or more, multilingual speech, dialects, a 15-second output, and flexible duration as on for 3.0 and off for 2.6. Text-to-video, image-to-video, start and end frames, and native audio are marked on both. When "Multi-Shot" is enabled, the model automatically plans the shot transitions, and this switch is a prerequisite for enabling "Custom Multi-Shot". Off, output stays a single shot. On, you can leave the camera plan to the model or click Custom Multi-Shot and set how many shots you want and how long each lasts. The stored guide stops mid-sentence in that section, so later clicks are not described. Subject lock uses multi-image references, or video references as elements, to hold characters, items, and the scene. Speech controls let you mark who is talking, in Chinese, English, Japanese, Korean, or Spanish, including dialects and a language switch inside one scene. Signs and new lettering are a separate claim, aimed at layouts such as ads. The price block is credits only: native audio at 12 credits per second for 1080p and 9 for 720p, 8 and 6 with audio off, and 2 per second for voice control. Generating a 5-second 3.0 Native Audio 1080p video costs 60 credits in the first example. Five seconds without native audio at 720p is 30 credits. Five seconds of native audio at 1080p with voice tone control is 70 credits. That block does not price 4K.
Official documentationMotion Control starts from a still. You upload a reference video, and the page says Kling will replicate the movement, facial expressions, and details with zero character distortion. The path is three steps: a character image plus a reference that shows the expressions, an optional Bind Face Subject lock fed by multi-angle photos or a short video, then a download the page calls high resolution. The FAQ is stricter than the marketing line. Motion Control currently supports one primary subject, and extra faces lose to the dominant one. Human, anime, and 3D characters are in scope when the style matches. A human reference on an animal still may degrade. Preferred inputs are clear, front-facing images whose facing direction matches the clip. Motion Control focuses on movement and expressions. For speech and lip-sync, use standard Video 3.0 generation with text prompts, not the reference audio. The page says 3.0 holds the face better than 2.6 across angles and longer motion. It does not show a credit rate for this tool, so the VIDEO 3.0 per-second table should not be assumed to bill it.
Official documentationText-to-video starts in a prompt. You pick a model and describe scenes, characters, actions, or style. A second step lets that prompt cover several scenes and also native voiceover and sound effects. Export settings on the page are 720p, 1080p, or 4K, a duration written as 3 - 15 seconds, and an aspect ratio of 16:9, 9:16, or 1:1. The feature copy adds 60fps playback depending on the plan, and says a recurring character should keep appearance, expression, and movement. When your prompt includes multiple scenes, Kling 3.0 can organize them into up to five connected shots. That ceiling is this page’s. The guide, separately, lets Custom Multi-Shot set the count and the durations. The pages do not say which rule wins if they conflict. Audio here is explicit: create videos with built-in dialogue, sound effects, and ambient audio in five languages: English, Chinese, Japanese, Korean, and Spanish, plus American, British, and Indian English. The FAQ repeats 720p on the free tier and 1080p or 4K on higher plans. A footer strip lists “Free tier available” and “Watermark-free membership” as separate labels, without naming which paid option drops a watermark.
Official documentationFollows the homepage fashion line: generate on-model showcases, styling content, and product videos from product images, model images, or brand assets. Feed the garment, the model, or the brand frame as the image, and keep the prompt on fit, color, and setting. Stay inside 3 to 15 seconds, and pick 9:16 only on the text-to-video export that lists it. Festival films and view counts on the homepage are not proof a product still will match them.
Another suggested workflow is a conversation that needs more than one angle. Enable Multi-Shot so the model plans transitions, then open Custom Multi-Shot when each shot needs its own length. Mark who speaks, and stay with the languages the guide lists. Keep the whole generation between 3 and 15 seconds. Put the lines in the prompt. The motion FAQ does not lift speech from a reference video. Read the credit line first: the 5-second native-audio 1080p example is 60 credits, and voice tone control on that setup is 70. This is not a clip the directory generated.
A third suggested workflow is Motion Control when the performance already exists. Upload one character still and one reference of the movement. Turn on Bind Face Subject and add front, side, and expression views if you want the lock the FAQ describes. Keep a single primary subject. Skip dialogue on this pass. Speech and lip-sync are sent back to Video 3.0 text prompts. Do not bill the render with the VIDEO 3.0 table unless a public page ties them together. Suggested only: no reference was uploaded here, and the face lock was not scored.
Use the studio on kling.ai, or the quickstart downloads for iOS, Android, macOS M Chip, macOS Intel Chip, and Windows. The pages do not say the apps include every desktop control.
Choose text to video, image to video, motion control, image generation, sound generation, or the VIDEO 3.0 guide. The index lists those names and does not merge them into one form.
On text to video, set resolution, the 3 - 15 second duration, and 16:9, 9:16, or 1:1. On VIDEO 3.0, enable Multi-Shot before Custom Multi-Shot. On Motion Control, add the reference and, if needed, Bind Face Subject.
For VIDEO 3.0, line the mode and resolution up with the shown 5-second cases of 60, 30, or 70 credits.
A named plan price was not verified on the kling.ai pages reviewed here. A membership card with a plan name and an amount was not verified in the text reviewed here. The text-to-video FAQ still states a free tier and a paid floor: If you want 1080p or 4K, a longer clip, or the ability to use your videos commercially, that is where the paid plans come in. They start at $0.59 a month. It also says to sign up, get monthly credits, and generate in 720p. Supported resolutions include 720p on the free tier, with 1080p and 4K available on higher plans. Free-tier use is described as typically personal, while paid plans may include commercial rights, and the FAQ points readers to the official terms. The same page lists “Free tier available” and “Watermark-free membership” without linking the watermark label to a plan. VIDEO 3.0 pricing stays in credits: 12 or 9 per second with native audio, 8 or 6 without, and 2 per second for voice control, with 5-second examples at 60, 30, and 70 credits. The developer pricing URL did not return a usable table.
Other sites includes different monthly amounts. They are not in the public official text, so they are left out. The $0.59 line is a FAQ sentence, not a checkout screen, and the credit examples are not converted into dollars.
You are comparing another generator and do not need reference-video motion.
Hailuo AI is another AI Video entry. This page does not assign it a price or a feature list. The documented KLING AI difference is the still-plus-reference Motion Control path and the per-second VIDEO 3.0 credit examples. Choose Hailuo AI when that path is not the job.
The text-to-video FAQ says to sign up, receive monthly credits, and generate in 720p. Supported resolutions include 720p on the free tier, with 1080p and 4K on higher plans. It does not show how many monthly credits that tier includes. Paid plans start at $0.59 a month when you want 1080p, 4K, a longer clip, or commercial use.
SourceCredits, not a currency amount. Native audio is 12 credits per second at 1080p and 9 at 720p. Without it, the rates are 8 and 6. Voice control is 2 credits per second at both sizes. The examples are 60 credits for 5 seconds of native audio at 1080p, 30 credits for 5 seconds without native audio at 720p, and 70 credits when voice tone control is added to the 1080p native-audio case. 4K is absent from that block.
SourceThe guide and the text-to-video page both run duration from 3 to 15 seconds. Text-to-video export also names 720p, 1080p, 4K, 16:9, 9:16, and 1:1, and it says playback can reach 60fps depending on settings. The guide’s credit table names only 1080p and 720p, so those rates are not a 4K price.
SourceNo on both counts. The FAQ says one primary subject, with extra faces dropped in favor of the dominant face. The tool covers movement and expressions. Speech and lip-sync go to Video 3.0 text prompts. VIDEO 3.0’s own table does mark multi-character coreference of three or more, which is a different route.
SourceNo. Named plan prices were not verified in the text reviewed here. The only dollar amount there is the FAQ line that paid plans start at $0.59 a month. The developer pricing page was not verified here, and outside price tables are not used.
SourceIt lists iOS, Android, macOS M Chip, macOS Intel Chip, and Windows, and it includes support@kling.ai. The footer also labels an API Platform. That label is not a rate card. Avatar 2.0, Element Library 3.0, and the Omni guides are names on the index only.
SourceOct 6, 2026 · NavHunt editorial
NavHunt did not run KLING AI. No prompt was submitted, no credit balance was spent, and no clip was downloaded. This profile only restates controls, durations, credit examples, and the single $0.59 sentence found on public kling.ai pages. This description is based on KLING AI's official materials reviewed on 2026-10-06, including KLING AI homepage, KLING AI quickstart index, Kling VIDEO 3.0 model guide, KLING AI Motion Control.
Match mode and resolution to the guide’s 5-second examples before assuming a monthly pool.