# Introducing WeatherNext 3, our most advanced and accurate global weather AI model

The model generates hourly forecasts at up to 5-kilometer resolution by training directly on live satellite mosaics and weather station observations.

- Powers forecasts across Google Search, Gemini, Google Maps, Maps Platform Weather API, and Google Earth Engine.
- Forecasts key surface variables at 5 km, other surface metrics at 10 km, and atmospheric variables like wind speed at 25 km.
- Adds specialized forecasts for clean energy, including 100-meter turbine-height wind speeds, cloud cover, and solar radiation levels.
- Improves precipitation Continuous Ranked Probability Scores by up to 60% against NASA IMERG baselines.
- Available for querying in BigQuery and Earth Engine or bulk download via Google Cloud Storage.

## Why it matters

Developers, grid operators, and Google users receive higher-resolution hourly weather forecasts and renewable energy predictions globally without running costly physical simulations.

## Sources

- [Google DeepMind: Introducing WeatherNext 3, our most advanced and accurate global weather AI model](https://deepmind.google/blog/introducing-weathernext-3-our-most-advanced-and-accurate-global-weather-ai-model)

---

Summarized by dstilled on 2026-09-03. https://dstilled.ai/story/e4c8627a-c3c4-4e61-a91d-4fa9617a4b4e