DeepMind Weather AI

Master GraphCast, GenCast, and FGN — the AI models that outperform traditional weather forecasting and are reshaping how we predict our planet's future.

Comprehensive resource for Research Engineers targeting DeepMind Sustainability

The Three Core Models

The evolution of ML-based weather prediction at DeepMind

GraphCast
Deterministic GNN • Science 2023
Parameters
36.7M
Resolution
0.25°
Timestep
6 hours
Inference
<60 sec

First ML model to outperform ECMWF HRES on 90% of verification targets. Uses multi-scale icosahedral mesh with 16 message-passing layers.

GenCast
Diffusion Ensemble • Nature 2024
Parameters
~60M
Resolution
0.25°
Timestep
12 hours
Ensemble
50+ members

Probabilistic forecasts via diffusion. Outperforms ECMWF ENS on 97.4% of targets. Superior tropical cyclone and extreme weather prediction.

FGN
Functional Generative • 2025
Parameters
180M
Resolution
0.25°
Timestep
6 hours
Speed
8× faster

Single-pass ensemble via 32D noise injection. Powers WeatherNext 2 in Google Search, Maps, and Pixel Weather.

Encode-Process-Decode

The shared architecture pattern powering all three models

1
Encoder

Lat-lon grid → Icosahedral mesh
721×1440 → 40,962 nodes

→
2
Processor

Message passing on multi-mesh
16-24 GNN/Transformer layers

→
3
Decoder

Mesh → Output grid
Predicts Δ from current state

// Autoregressive rollout for 10-day forecast
Xt → Model → Xt+1 → Model → Xt+2 → ... → Xt+40

Learning Modules

Everything you need to master DeepMind Weather AI

🧠

Graph Neural Networks

Message passing, aggregation functions, and how GNNs capture spatial relationships on Earth's spherical surface.

Learn more →
🌀

Diffusion Models

Score matching, denoising, and probabilistic generation. How GenCast creates ensemble forecasts.

Learn more →
🌐

Icosahedral Mesh

Why spherical representation matters. Multi-scale hierarchies from M0 (12 nodes) to M6 (40,962 nodes).

Learn more →
📊

Evaluation Metrics

CRPS, RMSE, ACC, Brier scores. How to measure probabilistic forecast quality and ensemble calibration.

Learn more →
⚡

JAX & Infrastructure

Training pipelines, TPU optimization, gradient checkpointing, and distributed systems for weather ML.

Learn more →
🎯

Interview Questions

12+ technical questions with detailed answer frameworks. System design and research scenarios.

Start prep →

Ready to dive deep?

The interactive study guide covers all three models, mathematical foundations, atmospheric science basics, and practical interview preparation.

Open Interactive Study Guide →