NVIDIA Launches Earth-2 Family of Open Models — the World’s First Fully Open, Accelerated Set of Models and Tools for AI Weather

2026-01-26 · Source: NVIDIA Blog · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics, Emerging Technologies & Innovation · Depth: Intermediate, long

Summary

NVIDIA has launched the Earth-2 family of open models, libraries, and frameworks, establishing the world's first fully open, accelerated AI software stack for weather and climate prediction. This initiative makes AI weather forecasting accessible globally, covering all stages from initial observation data processing to generating 15-day global forecasts or local storm predictions. The new open models include Earth-2 Medium Range, powered by the Atlas architecture for high-accuracy medium-range forecasts across 70+ variables; Earth-2 Nowcasting, using the StormScope generative AI architecture for kilometer-resolution, zero-to-six-hour local storm predictions; and Earth-2 Global Data Assimilation, powered by HealDA, which generates initial atmospheric conditions in seconds on GPUs. These models aim to significantly reduce computational time and costs compared to traditional physics-based supercomputing methods, enhancing forecasting accuracy and fostering collaboration among scientists, startups, and government agencies.

Key takeaway

For research scientists and developers working on weather and climate modeling, NVIDIA Earth-2 offers a complete, open-source AI pipeline that significantly reduces computational time and enhances forecasting accuracy. You should explore integrating these models, such as Earth-2 Medium Range and Nowcasting, into your workflows via NVIDIA Earth2Studio or Hugging Face to develop customized, high-resolution predictions and accelerate scientific breakthroughs.

Key insights

NVIDIA's Earth-2 provides an open, accelerated AI software stack for comprehensive weather and climate prediction.

Principles

• AI models reduce weather forecasting computational costs.
• Open-source AI accelerates innovation in weather prediction.
• Generative AI can outperform physics-based storm forecasting.

Method

The Earth-2 framework integrates models for data assimilation, nowcasting (0-6 hours), and medium-range forecasting (up to 15 days), enabling end-to-end AI weather simulation and customization.

In practice

• Use Earth-2 Medium Range for 15-day global forecasts.
• Apply Earth-2 Nowcasting for high-resolution local storm predictions.
• Integrate Earth-2 Global Data Assimilation for rapid initial conditions.

Topics

• NVIDIA Earth-2
• AI Weather Forecasting
• Generative AI
• Open Models
• Climate Prediction

Code references

• nvidia/clara
• NVIDIA/earth2studio

Best for: Research Scientist, AI Scientist, AI Engineer, Data Scientist

Related on AIssential

• How NVIDIA's Earth-2 uses AI to Accurately Predict Weather — NVIDIA Earth-2 provides an open, GPU-accelerated AI stack for faster, more accurate weather and climate forecasting.
• How Earth-2 Delivers Global AI Weather Forecasts on Demand — AI models can simulate atmospheric physics directly from observational data, outperforming traditional methods.
• The next chapter in flood resilience: Open sourcing Google’s hydrology framework — Open-sourcing Google's AI hydrology framework democratizes advanced flood forecasting, enabling local adaptation and global resilience.
• Protecting cities with AI-driven flash flood forecasting — AI-driven flash flood forecasting using news data significantly enhances urban climate resilience globally.
• Towards a world where no one is surprised by a natural disaster — AI-powered forecasting and real-time alerts significantly enhance global preparedness and response to natural disasters.
• Kaggle Solution Walkthroughs: LEAP - Atmospheric Physics using AI (ClimSim) with GreySnow — Machine learning emulators can significantly reduce computational costs in climate modeling.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by NVIDIA Blog.