Genie 4 is Google DeepMind’s new real-time world model for creators, game developers and AI researchers. Unlike video generators that hand you a finished clip, Genie 4 builds a world you walk through with your keyboard and steer with text as it runs: change the weather, add a character, set an event in motion. It generates native audio and live dialogue, simulates physics and natural human motion, and uses Gemini to keep scenes coherent. Research preview, waitlist open now.

Google is rolling out SynthID in the Gemini app, making it easy to verify whether an image was created or edited with Google AI. Just upload the picture and ask. Gemini checks for the watermark and gives a clear, quick response.

An Agent that Plays, Reasons, and Learns With You in Virtual 3D Worlds. Powered by Gemini, it goes beyond following basic instructions to think, understand, and take actions in interactive environments – meaning you can talk to it through text, voice, or even images.

WeatherNext 2 is Google DeepMind’s most advanced AI weather model, delivering forecasts up to 8× faster and at hourly resolution. Using a new Functional Generative Network, it generates hundreds of realistic scenarios to better capture uncertainty and extreme events. WeatherNext 2 now powers forecasts in Search, Gemini, Pixel Weather, and Google Maps, and is available in Earth Engine, BigQuery, and Vertex AI.