Kling AI launched in June 2024 as a text-to-video model available to early testers in China. A global web beta followed in July. Over the next two years, Kuaishou expanded Kling from short prompt-generated clips into a family of models capable of editing footage, controlling character motion, and generating synchronized sound.
The releases came quickly. In December 2025, Kling introduced its O1 multimodal video model and, just two days later, Video 2.6 with native audio. Model announcements, early access, wider availability, and developer API launches also fell on different dates. That pattern continued with Kling 4.0, from its limited September 2026 rollout to full availability in October.
This timeline follows Kling AI’s major releases from its first beta in June 2024 through Kling 4.0 in October 2026. You can check when a model or feature first appeared, how each generation developed, and when a release reached early testers, the public, or the developer API.
Last updated: October 15, 2026.
2026: Kling 4.0 and the 3.0 Model Family
| Date | Milestone |
|---|---|
| October 15, 2026 | Kling 4.0 reaches full availability. The full creative-studio model supports 30-second native video, up to 10 keyframes, 15 combined Omni Reference assets, stereo audio, and targeted editing. Kling also documents MCP and CLI integrations for automated creation. Creative-studio access and developer API packages have different availability. |
| September 30, 2026 | Kling 4.0’s full-model capabilities are announced. The announced model specifies 30-second native generation, 10 keyframes, 15 reference assets, stereo audio, and video editing. Kling schedules a wider October rollout and describes 10-bit HDR and two-minute video extension as upcoming features in this September announcement. |
| September 28, 2026 | Kling 4.0 Flash begins limited early access. Ultra Yearly subscribers are first in line for the faster variant, which supports text-to-video, image-to-video, and Omni Reference, with clips of 3–20 seconds at 720p. The redesigned creation experience also starts rolling out. |
| July 15, 2026 | The newer Kling API expands to more video models. Kling 3.0, 3.0 Omni, 3.0 Motion Control, 2.6, 2.6 Motion Control, and 2.5 Turbo gain access to the updated developer API, with batch task lookup and pagination. The legacy API continues to operate. |
| June 17, 2026 | Kling 3.0 Turbo launches through the API. The new variant handles text-to-video and image-to-video at 720p and 1080p. The accompanying 3.0 Omni API upgrade accepts reference video up to 15 seconds and introduces 4K output for Omni video generation. |
| April 23, 2026 | Native 4K video generation arrives for Kling VIDEO 3.0. The video model gains direct 4K generation, preserving fine details from high-resolution reference material. This is a video-output milestone, distinct from the 4K still-image capability introduced with Image 3.0 in February. |
| March 4, 2026 | Kling VIDEO 3.0 Motion Control reaches the API. Element binding improves character identity consistency during reference-video motion transfer, especially as head angles and facial expressions change. |
| February 25, 2026 | The Kling 3.0 models reach the developer API. VIDEO 3.0, VIDEO 3.0 Omni, IMAGE 3.0, and IMAGE 3.0 Omni receive API access. Upgraded Elements can use video to define a character and bind a voice to that character. |
| February 5, 2026 | The Kling 3.0 family launches with Ultra early access. VIDEO 3.0 and VIDEO 3.0 Omni introduce up to 15-second generation, native multilingual audio, and multi-shot storytelling. IMAGE 3.0 and IMAGE 3.0 Omni arrive with 2K and 4K still-image output. Public access is announced for a later stage. |
2025: Kling 2.x, O1, Motion Control, and Native Audio
| Date | Milestone |
|---|---|
| December 17, 2025 | Kling VIDEO 2.6 Motion Control begins rolling out. Reference footage can drive a character’s body movement, gestures, and facial performance, with clips up to 30 seconds under the applicable orientation settings. Motion Control reaches the developer API on December 22. |
| December 16, 2025 | Voice Control expands VIDEO 2.6. Creators can assign preset or customized voices to generated characters and reuse those voices in new clips. The API records the Voice Control upgrade on this date. |
| December 4, 2025 | Kling AI Avatar 2.0 introduces five-minute portrait animation. The updated avatar model turns a character image and speech or singing audio into longer performances with more expressive gestures and facial movements. |
| December 3, 2025 | Kling VIDEO 2.6 introduces native audio. Text and image prompts can generate video, dialogue, narration, sound effects, and ambience in one pass. At launch, voice generation supports Chinese and English, with clips up to 10 seconds. |
| December 1, 2025 | Kling VIDEO O1 introduces unified multimodal creation and editing. Text, images, video, and character references guide generation and targeted scene changes. IMAGE O1 launches in the accompanying image-model line. |
| October 15, 2025 | Kling Avatar becomes available through the API. The avatar endpoint creates a speaking or singing character video from an image and text or audio input. This predates the Avatar 2.0 upgrade in December. |
| September 23, 2025 | Kling 2.5 Turbo launches. Text-to-video and image-to-video improve prompt handling, motion, and style consistency. The 1080p five-second generation rate drops from 35 credits on the compared Kling 2.1 mode to 25 credits at this launch. |
| August 1, 2025 | Audio-generation capabilities expand in the API. Text-to-audio and video-to-audio functions become available, including sound generation for uploaded videos. These create or attach sound to video and predate VIDEO 2.6’s native joint audio-video generation. |
| May 26, 2025 | The Kling 2.1 lineup is introduced. Standard 720p and Professional 1080p modes launch for the non-Master image-to-video model. Kling 2.1 Master advances motion and prompt adherence. Non-Master text-to-video is not available at the initial launch. |
| April 15, 2025 | Kling 2.0 Master launches globally. The new video generation improves motion and prompt responsiveness, and multimodal editing can replace, insert, or remove scene content. Kolors 2.0 appears as the associated image model. |
| March 27, 2025 | AI Sounds and longer Kling 1.6 extensions arrive. Audio can be generated from text or existing video. The Extend feature continues a video in five-second increments, with total length reaching up to three minutes. |
| January 23, 2025 | Elements introduces multi-image references to Kling 1.6. Creators can reference one to four images of people, objects, animals, or scenes to direct relationships and maintain recognizable subjects in image-to-video generation. |
2024: Kling 1.0, Global Beta, and Early Video Models
| Date | Milestone |
|---|---|
| December 23, 2024 | Lip-sync reaches the Kling developer API. Eligible videos generated with VIDEO 1.0 or VIDEO 1.5 can receive synchronized mouth movements for speech or song audio. This is a later access milestone, not the launch of native audio generation. |
| December 19, 2024 | Kling 1.6 launches. The upgrade improves responses to movement, action timing, and camera descriptions, plus facial expression, lighting, color, and motion quality. |
| September 19, 2024 | Kling 1.5 introduces 1080p and Motion Brush. Professional mode improves image quality and motion. Motion Brush lets creators draw movement trajectories for elements in a scene. |
| July 25, 2024 | Kling opens global web beta access. International creators gain access through the English-language web portal after the earlier application-based test in China. |
| July 6, 2024 | Kling’s web portal and new video controls debut at WAIC. The release introduces start-and-end-frame and shot controls, alongside ten-second text-to-video generation. Image-to-video and extension had arrived in the preceding month. |
| June 6, 2024 | Kling 1.0 begins application-based testing in China. Kuaishou introduces the original text-to-video model through KuaiYing, emphasizing realistic motion and physical interactions. The company publishes its formal press announcement on June 10. |
FAQs
When did Kling 4.0 launch?
Kling 4.0 Flash entered limited early access for Ultra Yearly subscribers on September 28, 2026. The full model was introduced in detail on September 30 and marked “Full Available” on October 15, 2026. These dates refer to different stages of the Kling 4.0 rollout.
What is the difference between Kling 4.0 and Kling 4.0 Flash?
Flash is the faster, lighter model available first to a limited audience. It supports text-to-video, image-to-video, and Omni Reference at 720p with clips of 3–20 seconds. The full Kling 4.0 model supports up to 30-second native generation, start-and-end-frame control, as many as 10 keyframes, expanded multimodal references, and targeted editing. Flash has a narrower feature set.
Why do Kling AI launch dates and API dates differ?
Kling’s creative-studio releases and developer API rollouts follow different schedules. Kling 3.0 was announced with Ultra early access on February 5, 2026, while the 3.0 video and image models reached the official API on February 25. Model announcements, limited access, public access, and API integration represent different milestones.
How long can different Kling models generate video?
Duration depends on the operation and model. VIDEO 2.6 originally produced clips up to 10 seconds with native audio, and VIDEO 3.0 increased native video generation to 15 seconds. The full Kling 4.0 model supports up to 30 seconds of native video generation, whereas Flash supports up to 20 seconds. Video extension and Motion Control follow different length limits. The older Kling 1.6 extension feature could extend a video to three minutes through multiple generations.
