#

Key takeaways

• Google DeepMind and Google Research released WeatherNext 3.

• The model forecasts weather at 5 km resolution.

• It produces hourly forecasts.

• Google says it will feed weather information in Search, Google Maps, and Gemini.

• Google says it is available on Google cloud platforms.

• The report says rain evaluations improved by 60% over WeatherNext 2.

What Google announced

TechCrunch reported on September 3, 2026 that Google DeepMind and Google Research released WeatherNext 3. The source describes it as an AI weather-forecasting model. Google says the model will feed weather information in Search, Google Maps, and Gemini.

The report also says WeatherNext 3 will be available on Google cloud platforms. That makes the release relevant both as a product update and as a model deployment announcement. The source does not add more technical detail beyond the items above.

What the model does

According to the report, WeatherNext 3 can forecast at 5 km resolution. It also produces hourly forecasts. Those two details suggest a focus on finer-grained and more frequent weather output.

The source says Google improved rain evaluations by 60% over WeatherNext 2. That is the only performance comparison provided. It should be read as a claim from the report, not as an independent verification.

Why the release matters

Weather forecasting models matter because they shape how weather information is generated and delivered. In this case, Google is tying the model to consumer-facing products and cloud access. That combination can matter for users who rely on weather data in daily tools.

The report does not describe training data, model limits, or operational safeguards. It also does not explain how Google measures the 60% improvement. Readers should treat the release as a product and capability update, not a full technical audit.

Operational considerations

The source gives only a narrow view of the model. It confirms forecast resolution, hourly output, and product integration. It does not say how the model performs across all weather conditions or regions.

That means users and teams should avoid assuming universal accuracy from one reported metric. A model can improve one evaluation and still have gaps in other settings. The source does not provide enough detail to judge those gaps.

Morocco relevance

The source reports no Morocco-specific deployment or partnership. The conditional global lesson is simple: better weather forecasting tools can support planning where weather-sensitive decisions matter.

For readers, the main point is to watch how forecast quality, update frequency, and product access affect practical use. The source does not establish any local implementation details.

Bottom line

WeatherNext 3 is a notable AI weather model release because it combines finer resolution, hourly forecasts, and product integration. Google also says it improved rain evaluations over WeatherNext 2.

The report is useful, but limited. It tells us what Google announced and where the model will appear. It does not provide enough detail to assess broader reliability or deployment impact.