Gemini Omni 1.2 Flash now available on AI Gateway

Vercel's AI Gateway, a single API that routes requests to models from many providers, now offers Gemini Omni 1.2 Flash, Google's latest video-generation model. The changelog says developers can call it with the same Gateway API key they use for other models and without setting up a separate Google account. Requests go through Vercel's AI SDK functions generateText and streamText, which return the video as an MP4 file. With streamText the video still arrives only as a completed file, not frame by frame.

Through the Gateway, the model generates 3- to 15-second videos with sound from a text prompt, edits 3- to 15-second source clips using up to six reference images, and extends a video by 3 to 10 seconds per call up to 30 seconds in total. A follow-up call can also refine a clip the model already produced instead of starting over. Output comes in 360p, 720p, 1080p or 4K, in landscape 16:9 or vertical 9:16.

Vercel says the Gateway charges the provider's price with no markup and no platform fee on inference. Google's own announcement, which Intel covered separately, puts that price at $0.07 per second of 720p video, 30% below Gemini Omni 1.1, and says the 15-second maximum is up from 10 seconds. The changelog makes no claims about quality; the model's improvements are Google's description, shown through its own examples.