VercelNew Video Generation Capabilities
Alibaba's Wan models are now integrated into Vercel's AI Gateway, bringing advanced video generation capabilities to developers. The suite includes six model variants covering different use cases:
- Text-to-Video (t2v): Generate videos from text descriptions with support for stylized outputs
- Image-to-Video (i2v): Animate static images into moving videos
- Reference-to-Video (r2v): Extract character appearance and voice from reference material to generate new scenes with consistent styling
- Flash variants: Accelerated versions for faster iteration
Implementation and Access
Developers can access Wan models in two ways:
AI SDK 6: Generate videos programmatically using the new generateVideo function. Configuration supports specifying resolution (480p, 720p, 1080p), duration, and provider-specific options like reference URLs for character transfer.
AI Gateway Playground: Test models interactively without code, with options to compare providers, refine prompts, and download results directly.
Availability and Limitations
Video generation is currently in beta and available exclusively to Pro and Enterprise plan subscribers and paid AI Gateway users. The feature requires authentication and billing setup before use. Detailed documentation and quick-start guides are available in Vercel's official AI Gateway documentation.