Kling 4.0

3 models available

Provider: KLING

Updated: Recently

Kling 4.0 generates 3-30 second videos at 720p or 1080p with native audio in every clip. Text to Video writes a scene from a prompt; Image to Video follows a first frame, up to 8 middle keyframes and an optional last frame; Reference to Video combines up to 10 images, 5 videos, 7 reusable elements and 3 voices, referenced in the prompt with @ mentions.

Available Models

Kling 4.0 R2V

kling-4.0-r2v

Experience the power of Kling 4.0 R2V for your AI applications.

Kling 4.0 Image to Video

kling-4.0-i2v

Kling 4.0 Image to Video generates a 3-30 second clip at 720p or 1080p that passes through the frames you provide: a required first frame, up to 8 middle keyframes in chronological order, and an optional last frame. Mention frames in the prompt as @firstframe, @keyframeN and @last_frame. The aspect ratio follows the first frame; native audio is always included.

Kling 4.0 Text to Video

kling-4.0-t2v

Kling 4.0 Text to Video turns a prompt into a 3-30 second clip at 720p or 1080p, in 16:9, 9:16, 1:1 or 21:9, with native audio generated together with the picture.

Kling AI · Kling 4.0 · Released Oct 15, 2026New model

You call the shots.

Kling 4.0 is Kling AI's new flagship video model: up to 30 seconds in one generation, up to 10 keyframes, 15 mixed references and two-channel stereo sound. Direct the whole scene through one API, at 720p or 1080p.

Start creating API docs

Kling 4.0 · Official sample

16:9 · 720p preview · Stereo

28.0s · One generation, nothing stitched

Duration: 3–30 s
Resolution: 720p / 1080p
Keyframes: Up to 10
References: Up to 15
Audio: Stereo
Aspect ratio: 21:9 → 9:16

01 / 08Long take · Keyframes

Thirty seconds, on your beats.

Kling 4.0 renders up to 30 seconds in a single generation — one continuous cast, set and camera instead of stitched clips. Pin up to 10 keyframe images to the timeline and the story moves through them in order: an entrance, a reveal, a turn.

Keyframes · Official sample

16:9 · 5 keyframes in

The opening second shows the five keyframe images that went inClick a beat to jump to it

02 / 08Omni Reference

Fifteen references. One coherent take.

Cast the shot with images, videos and elements in a single request. Character sheets, storyboards, grids, wireframes and UI mock-ups define who and what is on screen; reference videos define how they move.

Omni Reference · Official sample

16:9 · 6 references in

The opening frames show the six references — three actors, a motorbike, a car and props — before the generated chase.

Images10

Videos5

Elements7

Per request15

Videos: 30 seconds combined. Elements: up to 3 built from video, and an element can carry a voice. Audio input is voice reference only.

Borrow the motion. Keep your cast.

A reference video can drive choreography, camera path or pacing while images decide who and what appears. White-model and depth-map videos work as references too.

Input videoKling 4.0

The reference fight

Used for its choreography and the timing of every hit. Two character designs went in as images.

New fighters, same fight

A new pair and a new arena, moving blow for blow with the reference.

03 / 08Video editing

Change one thing. Keep the rest.

Coming soon on APIPod

Bring footage — up to 5 clips, one primary, 30 seconds combined — and describe the change. Kling 4.0 edits expressions, movement, camera position, background or style while the rest of the scene holds still. Kling 4.0 editing is coming to APIPod soon; Kling 3.0 Omni Edit is available today.

Input videoKling 4.0

The original scene

A rider on a cyberpunk street. A felt-style reference went in as an image.

The same scene in felt

Every move and camera beat survives; only the material of the world changes.

Add or replace: Swap a character, product or set with reference images or elements. The blocking and timing of the original stay.
Remove: Take people or objects out of a shot and let the background fill in behind them.
Restyle: Re-render the same performance as felt, clay, anime or a new grade without touching the motion.

04 / 08Prompt understanding

One line, or a full brief.

Prompts can run to 8,000 tokens. Give Kling 4.0 a single line and it fills in the story, blocking and camera with a director's eye; give it a detailed brief and it follows it as written.

One-line prompt

16:9 · 15 s · Stereo

One line · 6 words

An old man ranting about AI.

The room, the performance and the rant itself were all Kling 4.0's call.

Long prompt

16:9 · 7 s · 1080p

Full brief · ~90 words

A continuous 7-second shot set in the UK. Two friends who have not seen each other for a long time meet at the corner of a quiet street at dusk, surrounded by a diverse crowd going about their daily lives. They smile, walk quickly toward each other, and share a warm embrace. Their facial expressions and body movements are natural. A handheld camera follows them with subtle camera shake. Soft, warm lighting with a nostalgic cinematic film look. Include natural footsteps, street ambience, background music, and English dialogue.

Every instruction lands: the handheld shake, the warm dusk light, the footsteps and the dialogue.

05 / 08Sound · Language

Multilingual · Official sample

16:9 · Dialogue · Stereo

Turn the sound on: the voices and the room tone are generated together with the picture.

It speaks your language.

Two-channel stereo audio is generated with the picture — dialogue, singing, effects and ambience — with tighter lip sync. Characters speak, and on-screen text renders, in nine-plus languages.

中文
English
日本語
한국어
Español
Português
Deutsch
Français
हिन्दी

Beijing · Taiwanese Mandarin · Northeastern · Sichuanese · Cantonese — American · British · Indian English

Two-channel stereo: Dialogue, music, effects and ambience, mixed in stereo and synced to the action.
Tighter lip sync: Voices stay matched to mouth movement, including sung vocals.
Legible on-screen text: Signs, titles, logos and emoji render clearly and stay readable as the camera moves.

06 / 08Spec sheet

The official numbers.

Specifications as published by Kling AI at launch, side by side with Kling 4.0 Flash — the faster model in the same family.

Capability	Kling 4.0	Kling 4.0 Flash
Text-to-video	Yes	Yes
Image-to-video (first frame)	Yes	Yes
First & last frame	Yes	—
Multi-keyframes	Up to 10 keyframe images	—
Omni Reference	Yes	Yes
Video editing	Coming soon on APIPod	—
Duration	3–30 s	3–20 s
Resolution	720p / 1080p	720p
Dynamic range	10-bit HDR announced (not on APIPod yet)Coming soon	8-bit SDR
Aspect ratio	21:9 / 16:9 / 1:1 / 9:16 / Auto	—
Audio	Two-channel stereo	—
Prompt length	Up to 8,000 tokens	—
Reference inputs	10 images · 5 videos · 7 elements · 15 total	—
Video extension	Repeated, up to 2 minutesComing soon	—

— = not listed in Kling AI's official Kling 4.0 Flash specification. Source: Kling AI release notes, October 2026.

Need fast drafts first? Explore Kling 4.0 Flash

07 / 08API

One endpoint, async by design.

Kling 4.0 runs on the same video endpoint as every other model on APIPod. Submit a task, poll its status, collect the MP4 — no SDK and no queue to manage.

POST /v1/videos/generations with model kling-4.0-t2v, kling-4.0-i2v or kling-4.0-r2v — a task_id comes back immediately.
GET /v1/videos/status/{task_id} until the status is completed, or pass a callback_url.
Download the video from result. The same key works for every model on APIPod.

Get an API key API docs

curl https://api.apipod.ai/v1/videos/generations \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kling-4.0-t2v",
    "prompt": "Handheld long take at dusk: two old friends meet on a London street and embrace",
    "duration": 30,
    "resolution": "1080p",
    "aspect_ratio": "21:9"
  }'

# {"code":200,"data":{"task_id":"<task_id>","status":"pending"}}

Field names follow APIPod's unified video schema. See the API docs for the full parameter list.

08 / 08FAQ

Questions, answered.

What is Kling 4.0?

Kling 4.0 is Kling AI's next-generation, natively multimodal video model from Kuaishou. It generates 3 to 30 seconds of video with two-channel stereo audio from text, images, keyframes or up to 15 mixed references. On APIPod it is available as kling-4.0-t2v, kling-4.0-i2v and kling-4.0-r2v; video editing is coming soon.

When was Kling 4.0 released?

Kling AI announced Kling 4.0 on September 27, 2026 and made it available to everyone on October 15, 2026. Kling 4.0 Flash opened to a limited group of users on September 28.

How long can a Kling 4.0 video be?

3 to 30 seconds per generation, rendered in one continuous pass rather than stitched clips. Kling AI has also announced repeated video extension up to 2 minutes, marked as coming soon.

How do Kling 4.0 keyframes work?

Besides first-frame and first-and-last-frame image-to-video, kling-4.0-i2v takes up to 8 middle keyframes in keyframe_urls — up to 10 frames in total, depending on duration. Each one pins a visual state — a pose, a scene change, a product reveal — and the model moves the story through them in order.

What can I use as references?

Omni Reference takes up to 15 assets per generation: up to 10 images, 5 videos (30 seconds combined) and 7 elements, of which up to 3 can be built from video. Images can be character sheets, storyboards, grids, wireframes or UI; videos can guide performance, motion, camera and pacing, including white-model and depth-map videos. Audio input is limited to voice references.

Can Kling 4.0 edit an existing video?

Kling AI's Kling 4.0 can: provide up to 5 clips — one of them the primary video — each 3 to 30 seconds long and 30 seconds combined, then describe the change: add, replace or remove elements, or change an expression, movement, camera position, background or style while the rest of the scene stays stable. Kling 4.0 editing is coming to APIPod soon; today you can edit a 3–15.5 second clip with kling-3.0-omni-edit.

Which resolutions and aspect ratios does Kling 4.0 support?

On APIPod, 720p or 1080p, in 21:9, 16:9, 1:1 or 9:16 (reference-to-video also supports Auto). Kling AI has announced 4K and 10-bit HDR output; they are not available on APIPod yet.

Does Kling 4.0 generate sound?

Yes — two-channel stereo with dialogue, singing, sound effects and ambience, and more accurate lip sync. Speech covers Chinese, English, Japanese, Korean, Spanish, Portuguese, German, French, Hindi and more, with accents such as Beijing, Taiwanese, Northeastern, Sichuanese and Cantonese Chinese, and American, British and Indian English.

Should I use Kling 4.0 or Kling 4.0 Flash?

Use Kling 4.0 for the take you ship: up to 30 seconds at 1080p, first and last frames, keyframes, Omni Reference and stereo audio. Kling 4.0 Flash, coming to APIPod soon, is for fast, better-value drafts: 3 to 20 seconds at 720p from text, a first frame or Omni Reference.

How do I call Kling 4.0 on APIPod?

Send POST /v1/videos/generations with model set to kling-4.0-t2v (text), kling-4.0-i2v (first, middle and last frames) or kling-4.0-r2v (Omni Reference), then poll GET /v1/videos/status/{task_id} — or pass a callback_url — and download the result. One API key covers Kling 4.0 and every other model on APIPod.

Your next shot is thirty seconds long.

Plan it with keyframes, cast it with references, cut it in one pass.

Start with Kling 4.0 Talk to sales