Gemini Omni — Google's unified multimodal model family for video

Today, Google released an AI model that can basically work with anything. We're talking text, images, video, audio. You get the point, but the question is, did they just change the game? Or did they copy China? I'm Gigachin. I spent years working as a venture capitalist in China, and I still do most of my business over there, so I can't help but notice this kind of thing. I'd like to know what you think. You probably know that Google IO 2026 kicked off today. That's Google's annual developer conference held in Mountain View, California. And the CEO dropped some big news. Namely, Gemini Omni, which is a completely separate next generation family of AI models. You've got Omni Flash, Omni Pro, Omni Ultra, covering everyone from the kid who wants to make a meme to the movie studio trying something bold and new. And I'm not gonna lie, I'm excited because it does a lot. Most familiar video models, like VO, C-Dance, Sora, RIP, are essentially just specialized diffusion transformers. This means they basically pattern match, not so heavy on the reasoning. But Google is introducing a model that you can actually have a conversation with. That's a big deal. Think of it as doing everything you want to do natively. It's a unified single model. Fragmented architecture is fairly typical for AI models today, but as discussed, Omni can do quite a lot. So unlike specialized video tools like Runway, Pika, Kling, this comes with the depth of an efficient large language model. We can kind of see…