Evidence-backed European AI register
ECMWF — European Centre for Medium-Range Weather Forecasts
Reviewed identity, roles, model listings, pricing and evidence for ECMWF — European Centre for Medium-Range Weather Forecasts.
Dataset compiled 2026-08-27 · evidence over inference
Reviewed identity
- Headquarters scope
- EU body or consortium
- Headquarters record
- Reading (UK) and Bonn (DE), with a data centre in Bologna (IT)
- Legal entity
- intergovernmental organisation established by convention
- Ownership
- state-owned (jointly, by its member states)
- Founded
- 1975
- Confidence
- high
- Last verified
- 2026-08-22
Evidence boundaries
Organisation listings identify a reviewed canonical record. Model listings show an association in the source register; they do not by themselves establish model creation, current API availability, hosting location or infrastructure ownership.
Model listings
- AIFS Single v1.1
- Type
- weather (graph neural network + sliding-window transformer)
- Parameters
- unknown
- Context
- unknown
- Languages
- n/a
- Licence
- open-source licence for weights and Anemoi training configs; forecast data under CC-BY-4.0
- Release
- operational from 2025-02-25
- AIFS Single v2
- Type
- weather
- Parameters
- unknown
- Context
- unknown
- Languages
- n/a
- Licence
- open-source weights
- Release
- 2026-05-12
- AIFS ENS (ensemble)
- Type
- weather
- Parameters
- unknown
- Context
- unknown
- Languages
- n/a
- Licence
- open
- Release
- 2025
- Anemoi framework
- Type
- tooling
- Parameters
- unknown
- Context
- unknown
- Languages
- n/a
- Licence
- open source
- Release
- unknown
Profile note
The most consequential European non-text foundation model that almost never appears in AI directories. ECMWF is an intergovernmental body that in February 2025 made a machine-learned model (AIFS) part of its operational forecast suite — the first major weather centre to do so — and publishes both the weights and the training configs openly under CC-BY-4.0. A rare case where the European public sector, not a startup, holds the frontier in a modality.