Kling AI 4.0 generates 30-second videos with stereo audio

Kling AI released version 4.0 this month with substantial upgrades aimed at professionals who need longer video sequences, better audio, and multilingual support. The new model extends generation length to 30 seconds in a single pass, expands reference control through an upgraded system that accepts up to 15 assets, and introduces stereo audio with lip-sync support across more than ten languages. The company also made Kling 4.0 accessible via MCP and CLI interfaces for teams building workflows with AI agents.

What changed

The company announced Kling 4.0 alongside a lighter Flash variant designed for speed and cost efficiency. The flagship model increases the prompt token limit to 8,000, giving users more space to describe scenes and narrative intent. Users can supply up to ten keyframe images to anchor scene transitions and character states throughout a clip.

The “Omni Reference” system is the primary feature addition. It accepts up to 15 reference assets in a single generation: up to 10 images, 5 videos with a combined maximum of 30 seconds, and 7 elements. According to the company, storyboards, wireframes, image grids, white-model videos, and depth-map videos all work as input references. This represents a significant expansion of creative control compared to earlier versions.

A new “Creative Recreation” feature allows users to take an existing video as a template and swap elements like products or characters while preserving other aspects of the shot. According to the company, individual components such as facial expressions, body movements, camera angles, or backgrounds can be modified independently without changing the rest of the clip.

Audio improvements include two-channel stereo support and expanded multilingual lip sync. The company states the system now handles German, English, Japanese, Korean, Spanish, and various dialects and accents. In-video text generation also works in multiple languages and maintains stability during camera movements.

Kling 4.0 Flash is the companion release, optimized for faster generation at lower cost but limited to 720p resolution and 20-second clips. The creation interface has been redesigned across both variants. The company lists 10-bit HDR output at 4K and 1080p, as well as a video extension function to stretch clips up to two minutes, as “coming soon.” API pricing has not yet been announced, with support listed as pending.

Why it matters

The 30-second limit has been a constraint for AI video tools. Longer individual clips reduce the number of cuts needed and allow creators to build more coherent narrative sequences without combining multiple generations. Professionals in marketing, training, and advertising often require continuity that single-shot tools cannot provide.

The reference system expansion gives art directors and storyboard artists a method to enforce visual consistency before generation. Supplying ten keyframes and fifteen total assets lets creators lock down specific shots, character states, and transitions rather than leaving those decisions to the model. This moves closer to directed video creation than purely text-driven approaches.

Stereo audio and multilingual lip sync matter for international distribution. Many AI video tools ship with mono audio and limited language support, requiring creators to add audio separately or work exclusively in English. Support for ten-plus languages with regional accents means teams can generate videos for global audiences without additional post-production.

The MCP and CLI access indicates Kling AI is positioning itself for automation and agent integration. Teams using AI agents to automate workflows can now call Kling 4.0 programmatically, enabling batch video generation and dynamic video creation as part of larger pipelines.

However, competitive positioning requires careful evaluation. The company published a comparison between Kling 4.0, Wan 3.0, and Seedance 2.5. According to the source material, Kling 4.0 holds its own but does not pull ahead, particularly in logical shot sequencing, physics accuracy, and natural-sounding dialogue. This suggests the pace of model improvement in AI video may be beginning to stabilize among current-generation tools.

What to test

Before adopting Kling 4.0 for production work, professionals should verify these claims:

• 30-second coherence: Generate 30-second clips across multiple domains such as product demonstrations, interviews, and narrative sequences. Monitor whether temporal consistency holds throughout or if quality degrades toward the end.

• Reference system reliability: Test whether supplying ten keyframes and fifteen mixed assets constrains generation as intended or whether the model ignores certain references. Assess whether the system behaves predictably when references conflict.

• Multilingual lip sync: Generate dialogue in at least three languages from your target markets and compare against native speaker expectations. Verify whether lip movements match phonetically and whether accents are recognizable.

• Stereo audio spatial placement: Confirm whether stereo audio creates genuine spatial effects or simply duplicates mono output across two channels. Test scenarios combining dialogue and background sound.

• Flash variant speed and cost trade-off: Time generation for equivalent 20-second clips on both Kling 4.0 and Kling 4.0 Flash. Once pricing is announced, compare cost per generation to assess the value proposition.

• Comparison with established tools: Run identical prompts and reference images through Synthesia and HeyGen, which serve established markets in professional video generation. Evaluate whether Kling 4.0 meets or exceeds their output quality, speed, or feature set.

The company’s comparison with Wan and Seedance should be treated as illustrative rather than definitive. Independent testing remains necessary to validate competitive positioning.

The conclusion

Kling 4.0 addresses practical constraints that have limited AI video adoption in professional settings. Longer clips, expanded reference control, stereo audio, and multilingual support are functional additions for production workflows. CLI and MCP access signals the company intends Kling 4.0 to integrate into larger creative and automation systems.

The comparison with Wan and Seedance suggests that differentiation through incremental improvement may be outpacing breakthrough innovation in this category. Kling 4.0 appears competitive but not decisively superior. API pricing remains unannounced, which will significantly influence adoption decisions.

Professionals should test Kling 4.0 on representative projects before committing resources, particularly if they currently use established tools. The relevant question is whether Kling 4.0’s specific feature set and output quality justify migration or parallel use for your workflow.

Related reading

• Apertus 1.5: 262K context, image and audio support
• Synthesia vs HeyGen: AI Avatar Video Creation
• Motive AI Agents automate fleet management operations