Google DeepMind and Google Research have unveiled WeatherNext 3, a new artificial intelligence model for weather forecasting that claims superior accuracy and resolution compared to existing models. According to Google, the model will be integrated into various Google products, including Search, Maps, and Gemini, and will also be accessible to users and researchers via Google Cloud platforms. Samier Merchant, a senior staff engineer at Google, stated that this model will power several core variables across Google’s products.

WeatherNext 3 has demonstrated the highest accuracy among leading AI weather models, outperforming those developed by Google, Microsoft, Nvidia, and the European Center for Medium-Range Weather Forecasting (ECMWF), as well as traditional forecasts from the U.S. National Weather Service and the ECMWF. The model addresses key weaknesses of previous AI forecasting models by providing predictions at a resolution of 5 km, showing a 60% improvement in rain evaluations over its predecessor, and offering hourly forecasts instead of the standard six-hourly predictions. These advancements result from the model’s larger parameter count, tailored decoder heads, and training on data from specific weather stations, enabling more granular predictions and evaluation against ground-truth data. Ferran Alet, a staff research scientist manager at DeepMind, emphasized that Google’s focus is on delivering useful weather information to users, as weather-related queries constitute a significant portion of user searches.

This is a summary. Read the original at techcrunch.com.