Google has unveiled Gemini Diffusion, a experimental research model by Google.
Google is always working on approaches to improve their AI models, including making them more efficient and performant in the world of AI. Built on the concept of diffusion—a technique already powering leading-edge image and video generation—this new model transforms random noise into highly relevant text or code.
Google released an experimental demo of Gemini Diffusion today. It shows Gemini Diffusion is significantly faster than any of Google’s previous models, while still matching their top-tier coding performance.
Also Read, Gemini Expands Beyond Smartphones, Bringing AI to All Android Devices
Google is working on a 2.5 Flash Lite, which approaches to lowering latency in all Google’s Gemini models.
If you’re eager to experience Gemini Diffusion early demo, then please sign up for the waitlist.