Google is taking weather forecasting to the next level with WeatherNext2, an advanced AI-driven model developed by Google DeepMind.
This new tool is designed for both researchers and everyday users, offering fast, high-resolution weather predictions.
Built on the GenCast model, WeatherNext2 promises faster and more accurate forecasts, along with broader accessibility for various Google platforms.
What Is WeatherNext2 and Where It’s Used
WeatherNext2 is integrated with platforms like Google Earth Engine, BigQuery, and Google Cloud’s Vertex AI through an early access program.
The model is also powering features on consumer platforms such as Google Search, Pixel Weather, Gemini, and Google Maps Weather API, providing detailed forecasts instantly.
This means users can get precise weather updates directly on tools they already use every day.
Fast, High-Resolution, and Highly Accurate
According to DeepMind, WeatherNext2 is 8 times faster than the previous model, delivering hourly forecast resolutions.
It can simulate hundreds of weather scenarios and provide updates in under a minute using just one Tensor Processing Unit.
The model uses a Functional Network architecture, allowing it to generate multiple realistic weather possibilities instead of just a single forecast.
In fact, WeatherNext2 has outperformed its predecessor, GenCast, in 99.9% of forecasts for up to 15 days, making it one of the most accurate AI-based weather models available.
