Google’s WeatherNext 3: An AI Forecast Engine Baked into Search and Maps

Google has introduced WeatherNext 3, its latest AI-based global weather model that takes a different route from traditional physics-heavy forecasting. Instead of relying mainly on numerical simulations, the system ingests massive amounts of real-time satellite imagery and other observational data, then uses AI to infer how weather systems will evolve.

The new version delivers hourly, high‑resolution predictions with roughly five times the spatial detail of previous iterations. Google highlights more accurate precipitation and wind forecasts as key upgrades, which could help sectors like agriculture, renewable energy planning, and disaster preparedness.

For everyday users, the biggest change is that WeatherNext 3 is being wired directly into products people already use: Google Search, the Gemini assistant, Google Maps, Google Maps Platform, and Google Cloud. That means the familiar weather cards in search results and maps will progressively shift to the new AI‑driven data without requiring any extra app. Google also stresses that improvements should be noticeable not only in major cities but in suburbs and rural or mountainous areas where traditional models have struggled with local microclimates.