Science & Data

Aurora

A foundation model for the whole Earth system

Open code / restricted weightsModel

What it is

Microsoft Research's foundation model for the Earth system, pretrained on large atmospheric datasets and fine-tuned into specialized versions for weather, air pollution, and ocean-wave forecasting. Aurora 1.5 extends it with more variables and hourly forecasts.

Why it's interesting

An open, pip-installable AI weather model in the GraphCast lineage that runs forecasts far faster and cheaper than numerical models — with air-quality and ocean-wave heads that reach beyond standard weather prediction.

Use cases

  • Medium- and high-resolution weather forecasting
  • Air-pollution prediction
  • Ocean-wave and climate research

Who it's for

Atmospheric and climate scientists, forecasting teams

Setup

Moderate. Python, pip install microsoft-aurora, a GPU, and ERA5-style input data

Limitations & cautions

As a neural forecaster it can be arbitrarily poor in edge conditions with no accuracy guarantees, and the commercial path routes through Microsoft.

Editorial takeaway

AI weather models are quietly rewriting a field that ran on supercomputers for fifty years. This is one of the few you can actually pip install.

Related & alternatives