Kling 4.0
Kling AI video model
Coming soon
Kling 4.0 generation is coming soon
This workspace will open for generation once Kling 4.0 is available on our platform. In the meantime, you can create videos with Kling 3 or Seedance 2.5.
Create now with
Kling 4.0 extends the Kling workflow with native clips up to 30 seconds, keyframe guidance and Omni references. Its published capabilities include output up to 4K with 10-bit HDR and generated stereo sound.
Kling 4.0 · Model overview
Kuaishou's Kling 4.0 offers a longer canvas for planning a scene: up to 30 seconds in one generation, 10 keyframes and 15 multimodal references. Use the timeline to define the action, references to establish the subject, and output options up to 4K 10-bit HDR for the intended finish.
A 30-second generation can carry an opening, a developing action and a resolved ending. That is twice the 15-second limit of Kling 3.0, but the brief still needs a deliberate pace.
Up to 10 keyframes can mark important compositions or story moments. Explain the action connecting them so the timeline communicates more than a series of unrelated pictures.
Omni references allow up to 10 images, 5 videos and 7 subjects within a combined limit of 15. Assign roles for identity, products, style and motion, including reference-led video revisions.
Plan fast movement with a clear path and camera relationship. The model targets greater stability in continuous action and complex shots; inspect the result for continuity throughout the movement.
Resolution reaches 4K, with 10-bit HDR available at 1080p and 4K. Choose the output for the editing and display workflow, especially when gradients and color grading matter.
Stereo sound can be generated with the picture. Identify speakers, languages and the timing of effects, then review speech and lip movement in the finished take.
Extension can continue a generated scene toward a total of 2 minutes. Preserve the subject and setting descriptions while giving each added segment a distinct next action.
Key Kling 4.0 specifications at a glance.
| Specification | Kling 4.0 |
|---|---|
| Developer | Kling AI (Kuaishou Technology) |
| Versions | Kling 4.0 (also called Kling 4) and Kling 4.0 Flash |
| Single generation | Up to 30 seconds, generated natively |
| Video extension | Up to 2 minutes in total |
| Keyframes | Up to 10 |
| Omni references | Up to 15: 10 images, 5 videos and 7 subjects |
| Prompt length | Up to 8,000 tokens |
| Resolution | Up to 4K |
| HDR | 10-bit HDR at 1080p and 4K |
| Aspect ratios | Adds 21:9 ultra-wide |
| Audio | Dual-channel stereo with more accurate lip sync |
| Languages | Chinese, English, Japanese, Korean, Spanish and more, with accents and dialects |
The full model emphasizes visual detail and control. Flash provides a speed- and cost-focused option; compare the available settings and estimate for the particular take you want to make.
Full model for high-fidelity scenes
Lightweight version for high-frequency creation
Use a 30-second timeline to test a complete narrative beat. Anchor essential compositions with keyframes and decide whether an extension serves the story after reviewing the first clip.
Build a product scene around authorized brand references. Output up to 4K HDR can suit a detailed campaign edit, but inspect product shape, labels and material before publication.
Plan short dialogue with a named speaker and language for each line. Review pronunciation and lip movement before preparing localized versions of an explainer.
Describe the path of a chase, sporting action or camera sweep. Keep spatial relationships clear and check the output for inconsistent movement or changing scene details.
Place up to 10 keyframes at the moments a storyboard needs to communicate. Use the generated sequence to discuss staging and pacing before final production.
Compose a wide scene at 21:9 and choose 10-bit HDR when the finishing workflow benefits from it. Review the result on an appropriate display before grading or delivery.
Creation workflow
Define the scene's progression before adding inputs. Keyframes establish important views, references establish appearance, and the prompt directs the movement and sound between those views.
Step 01
For a 30-second brief, give 0–10 seconds to establishing the subject, 10–20 seconds to the central action and 20–30 seconds to its resolution. Adjust that division to the event instead of forcing every scene into the same rhythm.
Step 02
Choose the key moments that need a reference composition, within the 10-keyframe limit. Add subject, image or video references for appearance and motion, keeping the combined reference set within 15.
Step 03
Write the action, camera direction, lighting and sound for each part of the sequence. The 8,000-token prompt capacity accommodates detail, but clear ordering matters more than filling the limit.
Step 04
Select the canvas, including 21:9 when suitable, and the required resolution. Review the first result before extending; continued segments can build a sequence up to 2 minutes.
Use the additional controls to remove ambiguity from the brief. Each keyframe, reference and sound instruction should explain something necessary about the scene.
Anchor the opening and ending compositions first. Add intermediate frames where a specific change of view or story event needs to be established.
Choose clear references with compatible appearances for each subject. Avoid mixing several contradictory designs for the same character or product.
Assign a camera move to each relevant part of the timeline. State how it relates to the subject instead of asking for several different moves at once.
Quote the dialogue and identify who says it, in which language or accent. Check the generated speech rather than assuming the wording and pronunciation will be exact.
Describe when ambience, effects and music should be heard. Keep the sound direction focused so it supports the main action and any spoken lines.
For an extension toward 2 minutes, retain the established subject and visual treatment. Change the next event while checking how the new segment joins the previous ending.
Compare the published output and reference capabilities below, then check the controls available in the workspace for the task you intend to submit.
| Capability | Kling 4.0 | Kling 3.0 | Seedance 2.5 |
|---|---|---|---|
| Single generation | Native 30 seconds | 3–15 seconds | Up to 30 seconds |
| Story control | Up to 10 keyframes for timeline-level control | Multi-shot storytelling from a single prompt | First-frame control and 3D spatial pre-visualization for camera blocking |
| References | Up to 15: 10 images, 5 videos and 7 subjects | Kling 3.0 Omni: up to 7 images and 1 reference video | Up to 50: 30 images, 10 videos and 10 audio files |
| Resolution | Up to 4K, with 10-bit HDR at 1080p and 4K | 720p and 1080p, plus a native 4K mode | 720p up to native 4K |
| Audio | Stereo audio, more accurate lip sync, multiple languages, accents and dialects | Native audio with multilingual speech and accents | Native audio, with audio files accepted as references |
| Longer sequences | Video extension up to 2 minutes in total | Multiple shots within a single clip | Extend or edit an attached video |
| Best fit | Scenes needing timeline keyframes, a defined reference set and an HDR finishing workflow | Cinematic clips of up to 15 seconds for social media, ads and short scenes | Reference-heavy 30-second productions and 4K brand content |
The Kling family began in June 2024 and has developed through several releases. This timeline highlights changes relevant to clip planning, reference input and output formats.
| Version | Release | Highlights |
|---|---|---|
| Kling 1.0 | June 2024 | Public launch of Kling text-to-video and image-to-video generation |
| Kling 1.5 | September 2024 | 1080p high-definition output and improved motion quality |
| Kling 2.0 | April 2025 | Expanded motion and prompt handling, visual improvements and multimodal editing capabilities |
| Kling 2.5 Turbo | September 2025 | Faster, more cost-efficient generation with stronger prompt adherence |
| Kling 2.6 | December 2025 | Introduced native generation of dialogue, effects and environmental sound in the Kling family |
| Kling 3.0 | February 2026 | 3–15 second clips, multi-shot storytelling and multilingual native audio |
| Kling 4.0 | 2026 | Up to 30 seconds per generation, 10 keyframes, 15 references, 4K 10-bit HDR and extensions toward 2 minutes |
These briefs give a 30-second sequence a visible progression and an explicit sound direction. Replace the subjects and references with materials suited to your own project.
0–10s: A white ceramic speaker sits on a shelf beside a closed book, soft morning side light, slow push-in. 10–20s: Move around the speaker as a hand turns its dial and a small indicator lights. 20–30s: Pull back to show the quiet living room, keeping the speaker's shape and finish consistent. Stereo room ambience and a gentle bass pulse; no dialogue.
A small bakery before opening. A woman in a blue apron places bread on the counter and says in English, "The first batch is ready." Her colleague looks up and answers in French, "Ça sent bon." Slow lateral camera move, warm oven light and quiet kitchen ambience. Keep each voice with its speaker.
Use keyframes 1, 5 and 10 for the beginning, middle and ending: a musician carries a brown guitar case out of a station, walks through a quiet square and reaches a theater door. Preserve the coat and case. Follow at waist height in 21:9, with footsteps and a gradually rising acoustic rhythm.
0–10s: A mountain biker approaches a bend on a dry hillside trail, low tracking camera. 10–20s: The rider follows the curve while the camera rises to reveal the valley. 20–30s: The rider slows beside a wooden lookout and puts one foot down. Keep the bicycle and helmet consistent. Stereo tire noise, wind and distant birds.
A presenter in a softly lit studio holds a compact desk lamp and says in English, "A little light for your reading corner." She turns the lamp toward an open book and pauses. Medium framing, visible hand movement, quiet room ambience. Keep the product shape consistent and add no unsupported specifications.
21:9, 10-bit HDR. 0–10s: Glide slowly above a misty lake at dawn. 10–20s: Descend toward a small wooden boat near the reeds. 20–30s: The boat drifts into a band of sunlight as the camera comes to rest. Natural reflections, restrained color, soft water and a low orchestral note.
Guidance on timeline planning, input limits, output formats and the differences between Kling versions.
Kling 4.0 is Kuaishou's fourth-generation Kling video model. Its published features include native clips up to 30 seconds, 10 keyframes, 15 references, stereo audio and output up to 4K with 10-bit HDR.
The workspace displays the current generation availability. When submitting is enabled, requests use credits and show an estimate before confirmation. Preparing a scene brief and reading the model guide do not require a model download.
The full Kling 4.0 model targets detailed scenes and creative control, with 30-second generation, 10 keyframes, 15 references and up to 4K 10-bit HDR. Flash is the speed- and cost-focused variant. Check the actual settings and quote for the selected option.
One generation can reach 30 seconds, compared with 15 seconds for Kling 3.0. Extension can continue a clip through additional segments to a total sequence length of up to 2 minutes.
The published output range reaches 4K. 10-bit HDR is supported at 1080p and 4K, providing more tonal steps than 8-bit video for compatible grading and display workflows.
Omni Reference permits up to 10 images, 5 videos and 7 subjects, but the combined reference count cannot exceed 15. Keyframe control separately supports up to 10 timeline anchors.
Omni Reference combines visual inputs to guide identity, style and motion. Its limits are 10 images, 5 videos and 7 subjects within 15 references total; explain the role of each asset in the brief.
Kling's published Omni update includes enhanced video editing. Up to 5 video references can inform a revision; use the prompt to distinguish what should change from the content that should remain.
Kling 4.0 adds 21:9 for wider compositions alongside the familiar landscape, portrait and square formats. Choose the ratio that suits the scene's destination.
The model targets improved handling of fast action, continuous movement and complex camera paths. Describe the action clearly and inspect the whole clip for spatial or motion inconsistencies.
Stereo sound is generated with the picture. Supported speech languages include Chinese, English, Japanese, Korean, Spanish, Portuguese, German, French and Hindi, with accent and dialect options. Check voice assignment, pronunciation and synchronization in the result.
Compared with Kling 3.0, the published 4.0 capabilities extend single clips from 15 to 30 seconds and references from 7 images plus 1 video to a combined 15 multimodal references. Other additions include 10 keyframes, 10-bit HDR, 21:9 output, stereo sound, prompts up to 8,000 tokens and extension toward 2 minutes.
Both list clips up to 30 seconds and output up to 4K. Kling 4.0 offers timeline keyframes, 10-bit HDR and extension toward 2 minutes. Seedance 2.5 allows up to 50 references, including audio, for briefs built from a larger set of source materials.
The prompt capacity is 8,000 tokens. Use the space to organize the scene by shot or time segment, including the relevant camera, dialogue and sound instructions.
Write the beginning, central action and ending as separate beats. For each, describe the subject movement, camera and light. Quote dialogue and name the speaker and language, then add timed sound cues. Use references and keyframes to clarify appearance and composition; text should explain what happens between them.
Kuaishou Technology develops Kling through its Kling AI video platform. The family began with Kling 1.0 in June 2024 and has since added successive model releases.