A groundbreaking artificial intelligence model developed by Google offers weather forecasts with unmatched accuracy for up to 15 days, potentially saving lives as climate disasters escalate, the tech giant announced.

The model, called GenCast, was created by Google’s DeepMind, its AI research lab based in London. According to Google, GenCast demonstrates superior predictive capabilities compared to today’s leading reference model, the European Centre for Medium-Range Weather Forecasts (ECMWF).

The ECMWF currently provides forecasts for 35 countries and is widely regarded as the global benchmark for meteorological accuracy. However, GenCast outperformed ECMWF’s forecasts in over 97% of the 1,320 climate disasters recorded in 2019, as per Google’s analysis.

Trained on historical data spanning four decades (1979–2018), including temperature, wind speed, and atmospheric pressure, GenCast can produce 15-day forecasts in just eight minutes—a process that traditionally takes hours. “GenCast provides more accurate predictions for both daily weather and extreme events, extending up to 15 days in advance,” Google stated in its announcement.

Google emphasized that GenCast consistently surpassed the existing reference model in forecasting extreme heat, severe cold, and intense wind speeds. “More precise predictions of extreme weather risks can help authorities protect more lives, prevent substantial property damage, and save money,” the company underscored.