Kling AI releases 4.0: 30-second video, stereo audio, 10+ language lip sync, plus a 720p Flash tier

Kling AI released Kling 4.0 on October 19, 2026. The video generation model produces up to 30 seconds of video in a single pass, adds two-channel stereo audio, and supports lip sync across more than ten languages, according to the company. A lighter variant, Kling 4.0 Flash, ships alongside it, limited to 720p resolution and 20 seconds of video.

Key specifications

• Maximum clip length: 30 seconds per generation (Kling 4.0); 20 seconds (Kling 4.0 Flash)
• Flash resolution cap: 720p
• Keyframe inputs: up to 10 images to define scene transitions, character states, and story beats
• Prompt comprehension: expanded to 8,000 tokens
• Audio: two-channel stereo
• Lip sync: more than 10 languages, including German, English, Japanese, Korean, and Spanish, plus various dialects and accents, according to Kling AI
• API pricing: not yet disclosed; API support is listed as "coming soon"
• Parameter count and training data cutoff: not disclosed

Reference and editing controls

The upgraded "Omni Reference" system accepts up to 15 reference assets per generation, according to Kling AI: up to 10 images, 5 videos with a combined maximum of 30 seconds, and 7 elements. Storyboards, wireframes, image grids, white-model videos, and depth-map videos can all serve as references.

A "Creative Recreation" feature lets users take existing videos as templates and swap out products or characters. Kling AI says individual elements such as facial expressions, body movements, camera angles, or backgrounds can be changed without affecting the rest of the clip. In-video text generation now works in multiple languages and, the company claims, stays stable during camera movements.

Access and availability

The creation interface has been redesigned. Kling 4.0 can also be accessed through MCP and CLI, which allows AI agents to drive automated video workflows. Two features are listed as "coming soon": 10-bit HDR output at 4K and 1080p, and a video extension function that would stretch clips to up to two minutes. No release dates were given for either.

Competitive context

Kling AI says the release should close the gap between AI-generated and professional video. The Decoder's own side-by-side comparison against Wan 3.0 and Seedance 2.5, both from Chinese competitors, found that Kling 4.0 holds its own but does not pull ahead. The gaps were most visible in logical shot sequencing, physics, and natural-sounding dialogue. No standardized benchmark scores were published with the release.

What this means

The most consequential changes in Kling 4.0 are about control and integration rather than raw output quality. Ten keyframes, 15 mixed-media references, and element-level edits target the main production problem with AI video, which is consistency across shots. MCP and CLI access signals that Kling expects agents, not just people in a web interface, to be a meaningful share of usage.

The Flash tier follows a common pattern: a cheaper, faster model with capped resolution and duration for iteration, and the full model for final output. Without published pricing, it is impossible to judge how much cheaper Flash is or whether the API is competitive. Teams planning to build on Kling programmatically will have to wait for that information.

The Decoder's assessment that Kling 4.0 does not clearly beat Wan 3.0 and Seedance 2.5 suggests differentiation is shifting from base model quality toward features, audio, and workflow tooling. That is one comparison, not a benchmark, so it is too early to call a plateau. The next Wan and Seedance releases will test that reading.