4 September 2026

Google DeepMind releases WeatherNext 3 weather forecasting model

First reported

Engadget ran this on , a day before the other 3 sources picked it up.

  • WeatherNext 3 uses live satellite data instead of older numerical models, eliminating a six-hour lag and updating forecasts every hour.
  • The model generates forecasts on a 5-kilometer grid instead of 25-kilometer, showing five times more temperature detail and reducing rain prediction errors by up to 50 percent.
  • WeatherNext 3 predicts wind speeds at turbine height and solar radiation levels, helping renewable energy companies optimize power generation.
  • The model is now integrated into Google Search, Maps, Gemini, and Earth, with a public version available in Google Weather Lab.

Where they differ

All three newsletters agreed on the core facts.

  • The Rundown AI

    emphasized this as an underrecognized AI application in everyday tools.

  • Deep Learning Weekly

    focused more on the technical improvements.

What each one reported

TLDR AITLDR editorial team

WeatherNext 3 is Google DeepMind's most accurate global weather model, adding real-time satellite data, hourly refreshes, higher resolution, precise precipitation forecasting, and clean energy variables.

The Rundown AIRowan Cheung

Google DeepMind introduced WeatherNext 3, which achieves 5x improvement in temperature detail and reduces rain forecast errors by up to 60% through hourly satellite image updates. The newsletter emphasizes this as a major AI win that most people don't recognize as AI, now integrated across Google Search, Maps, Gemini, and Earth.

Deep Learning WeeklyEditorial team

WeatherNext 3 delivers 5km hourly forecasts five times sharper than its predecessor, improving precipitation accuracy up to 60% against NASA satellite data.

Reported by Engadget