SHELFCRITTER AI Digest

MODEL

WeatherNext 2

modeltopic-notedeepmindgoogleopen-source

Overview

WeatherNext 2 is DeepMind‘s general-purpose weather forecasting model, open-sourced Aug 6, 2026 alongside two sibling models — WeatherNext Cyclones (specialised for tropical-cyclone tracking) and WeatherNext 2-mini (small enough to run inference on a single TPU in Google Colab). The release lands alongside a Nature paper on cyclone forecasting that claims a roughly full-day lead-time advantage over operational cyclone models in current use by national weather services. Fits DeepMind‘s AlphaFold / GraphCast narrow-science outreach pattern rather than a shift on frontier-model openness.

Timeline

  • 2026-08-07-AI-Digest — DeepMind on Aug 6 open-sources three variants — WeatherNext Cyclones, WeatherNext 2 (general-purpose), and WeatherNext 2-mini (single-TPU Colab inference) — alongside a Nature paper on cyclone forecasting claiming a roughly full-day lead-time advantage over operational cyclone models. Not a frontier-lab openness moment — weather forecasting is a narrow, non-agentic, non-conversational scientific domain, and no frontier-lab weights (Gemini 3 Pro, Gemma 4 family) are being released.

Key Developments

  1. Open-Source Release Alongside Nature Cyclone Paper (August 6, 2026): Three-variant open-source release (WeatherNext Cyclones + WeatherNext 2 + WeatherNext 2-mini) alongside a Nature paper on tropical-cyclone forecasting claiming a full-day lead-time advantage over operational cyclone models used by national weather services. Fits the DeepMind outreach-and-partnership-with-domain-institutions playbook (AlphaFold, GraphCast, MedGemma), not the open-vs-closed frontier debate — designed to be consumed by national weather services and academic groups that lack the training compute for foundation-scale forecasting models. The single-TPU-in-Colab framing is genuinely useful: it lets domain scientists run experiments without a GPU-cluster procurement cycle. 30/60/90-day watch: whether national weather services (NOAA, ECMWF, JMA) integrate WeatherNext 2 into operational pipelines or keep it as a research reference — that is the practical impact test, not download counts.

See also: DeepMind, Google, MOC - Open Source Models.