Kling Video Models Now Available in AI Gateway
Vercel has integrated Kling's video generation models into AI Gateway, enabling developers to generate cinematic videos programmatically. The integration includes multiple Kling model versions, from the latest Kling 3.0 to the faster Kling 2.5-turbo variants, with support for both text-to-video and image-to-video generation.
Key Capabilities
The available Kling models support several advanced features:
- Text-to-Video Generation: Create videos from natural language prompts with the v3.0 models offering the highest quality
- Image-to-Video Animation: Animate static images with coherent motion and physics
- Multishot Narratives: Kling 3.0 can generate multi-scene videos from a single narrative prompt with intelligent shot composition
- First & Last Frame Control: Specify both start and end frames to control precise scene transitions and create time-lapse effects
- Audio Generation: Generate synchronized sound effects and ambient audio alongside video (v2.6 models)
- High Resolution Output: Support for up to 1080p generation in pro mode
Getting Started
Two approaches are available for developers:
AI SDK 6: Use the
generateVideo()function to generate videos programmatically with full control over models, prompts, duration, aspect ratio, and provider-specific options.AI Gateway Playground: Experiment with video models through a no-code interface embedded on each model page, allowing you to compare providers, tweak prompts, and download results without writing code.
Availability and Pricing
Video generation is currently in beta and available to Pro and Enterprise plan subscribers and paid AI Gateway users. The feature is accessible through both the AI SDK and the interactive Gateway Playground for rapid experimentation.
Available Models
Six Kling models are supported:
klingai/kling-v3.0-t2vandklingai/kling-v3.0-i2v(latest generation with multishot)klingai/kling-v2.6-t2vandklingai/kling-v2.6-i2v(audio generation support)klingai/kling-v2.5-turbo-t2vandklingai/kling-v2.5-turbo-i2v(faster generation)