Google has unveiled WeatherNext 3, its latest AI-powered weather forecasting model developed by Google DeepMind and Google Research. This new system significantly advances meteorological prediction by providing more frequent updates and precise readings of crucial variables such as temperature, windspeed, humidity, and rainfall. WeatherNext 3 produces hourly forecasts at a high resolution of 5 kilometers, an improvement from previous models, and it integrates its data directly into Google Search, Google Maps, and Gemini, as well as making it accessible to researchers via Google’s cloud services.

Compared against other state-of-the-art AI models from organizations like Microsoft, Nvidia, and the European Center for Medium-Range Weather Forecasting (ECMWF), Google’s WeatherNext 3 has shown superior accuracy on the Operational WeatherBench testing platform. It also outperforms traditional government forecasts, such as those from the U.S. National Weather Service and the ECMWF. This breakthrough is due to the model’s larger size, with 2.4 times more parameters than its predecessor, and its design that tails forecast outputs to specific locations, providing more practical and ground-truth-verifiable insights.

The model stands out by directly assimilating raw satellite observations collected in real-time, a task that has been technically challenging but critical for improved accuracy and frequency in forecasts. While many forecasting systems rely on processed data from government supercomputers, WeatherNext 3’s ability to work with raw empirical data makes it the first AI model to offer such detailed global predictions. Though Google claims this first-of-its-kind approach, some startups like WindBorne have been utilizing raw data in their models with a narrower focus, highlighting the growing competition to refine AI weather forecasting further.

Google sees WeatherNext 3 as a crucial tool for delivering timely and reliable weather information to users worldwide, with applications that extend beyond daily convenience to impactful sectors such as renewable energy and agriculture. The model's enhanced resolution and more frequent updates could support better planning for wind and solar projects, ultimately aiding goals like improved crop yields in developing regions. As AI continues transforming meteorology, this development underscores Google’s commitment to integrating cutting-edge technology into everyday products, benefiting both consumers and researchers.