AI Potluck
Contribute
Back to Gap Map Model components / Scientific AI models

AIFS

ECMWF
open weights / Overall score: 1.7

AIFS is the European Centre for Medium-Range Weather Forecasts' data-driven forecasting system, a graph-neural-network encoder and decoder around a sliding-window transformer processor. AIFS Single v1.0 is the first operationally supported version, runs four times daily beside ECMWF's physics-based model, and ships its checkpoint alongside the Anemoi pretraining and fine-tuning configurations that produced it.

Openness

3 high confidence
3.0
weights
open(CC-BY-4.0 ungated aifs-single-mse-1.0.ckpt)
data
documented-not-released(ERA5 1979-2022 for pretraining and operational IFS analyses 2016-2022 for fine-tuning, with the full input and output field list
code
partial(config_pretraining.yaml and config_finetuning.yaml ship beside the checkpoint with pinned Anemoi versions and both stage commands, but no data-preparation recipe - the commands begin DATASETS_PATH=??????? and no anemoi-datasets build ships)
license
CC-BY-4.0(permissive non-OSI)

ECMWF publishes both the pretraining and fine-tuning configs beside the checkpoint, with pinned library versions and the commands to run them. But both commands point to an undefined dataset path, and no script builds the ERA5 and operational-analysis inputs the checkpoint was trained on, so this reads as open weights with partial code rather than a fully open release.

  • https://huggingface.co/api/models/ecmwf/aifs-single-1.0 recorded 2026-09-10

    license: cc-by-4.0; gated: false; private: false; siblings include aifs-single-mse-1.0.ckpt, config_pretraining.yaml and config_finetuning.yaml.

  • https://huggingface.co/ecmwf/aifs-single-1.0/raw/main/README.md recorded 2026-09-10

    "These model weights are published under a Creative Commons Attribution 4.0 International (CC BY 4.0)" while notebooks and scripts are Apache 2.0; gives the pretraining command anemoi-training train --config-name=config_pretraining.yaml against pinned anemoi-training 0.3.1, anemoi-models 0.4.0 and anemoi-graphs 0.4.4, then the fine-tuning stage; and details pretraining on ERA5 1979-2022 followed by rollout on operational IFS analyses 2016-2022.

Adoption

1 high confidence
1.0

Adoption is measured by Hugging Face downloads of the single published checkpoint over the trailing month, the only usage channel on record for AIFS.

Capability

2 high confidence
2.0

One attribute. The MSE-trained model predicts a fixed forecast target rather than generating, thirteen pressure levels are vertical structure rather than a second scale, and ERA5 with operational IFS analyses is one data system.

Verified 2026-09-10