AI video generator Kling AI has launched version 4.0, introducing the ability to generate up to 30 seconds of video in a single pass while processing up to ten keyframe images for precise scene and character transitions. The upgraded Omni Reference system accepts up to 15 reference assets, including depth maps and storyboards, while prompt comprehension has expanded to 8,000 tokens. Kling 4.0 also adds two-channel stereo audio, multilingual in-video text, and lip-sync capabilities across more than ten languages.
Alongside the full model, the company launched Kling 4.0 Flash, a faster and lower-cost variant limited to 720p output and 20-second clips. Access is now supported via Model Context Protocol (MCP) and Command Line Interface (CLI) to integrate into agentic workflows. Features like 10-bit 4K HDR export, video extension to two minutes, and API pricing remain listed as coming soon.
Why it matters
Multi-asset keyframing and reference controls reduce reliance on random generation for video creators.
MCP and CLI support enable developers to integrate multi-modal video generation directly into automated agent workflows.
Lighter variants like Flash optimize operational costs for high-throughput commercial media generation.
Source: the-decoder.com



