AI Gateway adds support for video generation

AI Gateway has started supporting video generation. Now, through AI SDK 6, you can programmatically generate high-quality, cinematic videos in addition to text and images.

Without writing any code, users can experiment with various video models in the AI Gateway Playground, adjust prompts, and download results instantly.

The major models currently supported are as follows:

• Grok Imagine (xAI): Fast speed and excellent instruction-following ability, supports style transformation.
• Wan (Alibaba): Specializes in reference-based generation and multi-shot storytelling, maintaining character consistency.
• Kling: Strengths in image-to-video and native audio, supports automatic scene transitions.
• Veo (Google): High visual fidelity and physics implementation, cinematic lighting and audio generation.

Video generation methods are broadly divided into four types:

• Text-to-video: Generates video by describing a scene with a text prompt.
• Image-to-video: Converts a still image into a moving video.
• First and last frame: Generates the scene between specified start and end frames.
• Reference-to-video: Extracts a character from a reference image or video and places it in a new scene.