A new artificial intelligence weather forecasting model developed by Google DeepMind, named GraphCast, has demonstrated superior accuracy and speed compared to the world’s leading traditional numerical weather prediction system. The AI model, described in a paper published in Science, can generate a 10-day global weather forecast in under one minute on a single Google TPU v4 cloud computer, a task that takes the European Centre for Medium-Range Weather Forecasts (ECMWF) system hours on a supercomputer. GraphCast was trained on nearly four decades of historical weather data from ECMWF. In a head-to-head comparison, it outperformed the ECMWF’s High RESolution (HRES) system on over 90% of 1,380 test variables, including critical predictions for severe weather events like tropical cyclones and atmospheric rivers. The model excels at tracking cyclone movement and predicting extreme temperatures. Researchers emphasize that AI models like GraphCast are not replacements but powerful complements to traditional physics-based systems, potentially enabling more frequent and detailed forecasts to improve early warnings for hazardous weather. The full research and details are available in the Science article: https://www.science.org/doi/10.1126/science.adl6127.