Back to Gap Map Infrastructure / Classic ML & computer vision

GluonTS

open source / Overall score: 3.0

Probabilistic time-series forecasting library from AWS, focused on deep-learning models in PyTorch with older MXNet implementations. It ships estimators such as DeepAR, Temporal Fusion Transformer, PatchTST, TiDE, N-BEATS, WaveNet and DeepVAR, plus tree-based and statistical wrappers, along with dataset loaders and evaluation tools. AWS's pretrained Chronos forecasters live in a separate repository.

Openness

5 high confidence

5.0

license: Apache-2.0(OSI)
source: public(github.com/awslabs/gluonts)
core features withheld: no — no paid tier or commercial edition found

Apache-2.0 and built from the public repository under the awslabs account. AWS sells no edition of it; its pretrained Chronos models are a separate open project.

https://pypi.org/pypi/gluonts/json recorded 2026-09-27
PyPI record for gluonts 0.17.0, uploaded 2026-07-31, license Apache-2.0, Repository URL github.com/awslabs/gluonts.
https://raw.githubusercontent.com/awslabs/gluonts/dev/LICENSE recorded 2026-09-27
LICENSE is the "Apache License, Version 2.0, January 2004".
https://raw.githubusercontent.com/awslabs/gluonts/dev/NOTICE recorded 2026-09-27
NOTICE: "Gluon Time Series / Copyright 2018-2019 Amazon.com, Inc. or its affiliates. All Rights Reserved."
https://raw.githubusercontent.com/awslabs/gluonts/dev/README.md recorded 2026-09-27
README: "GluonTS is a Python package for probabilistic time series modeling, focusing on deep learning based models, based on PyTorch."; points to Chronos in its own repository; no paid tier mentioned.
https://ungh.cc/repos/awslabs/gluonts/files/dev recorded 2026-09-27
Repository tree of 1,016 paths with no ee, enterprise, commercial or pro directory.

Adoption

3 high confidence

3.0

Measured on monthly PyPI downloads of the gluonts package.

https://pypistats.org/api/packages/gluonts/recent recorded 2026-09-27
last_month 279,795 downloads of gluonts

Capability

3 high confidence

3.0

GluonTS is a catalog of forecasting models with the tooling to train and compare them, which is what sktime offers too, though GluonTS concentrates on deep probabilistic models. Choosing among them and tuning them is the user's job, and its pretrained zero-shot models ship separately as Chronos.

https://ts.gluon.ai/stable/getting_started/models.html recorded 2026-09-27
"Available models" table of 25 rows including DeepAR, DeepState, DeepFactor, Deep Renewal Processes, GPForecaster, MQ-CNN, MQ-RNN, N-BEATS, Rotbaum, Temporal Fusion Transformer, Transformer, WaveNet, DeepVAR, GPVAR, LSTNet and wrappers for R forecast and Prophet.
https://ungh.cc/repos/awslabs/gluonts/files/dev recorded 2026-09-27
Tree under src/gluonts/torch/model/ includes d_linear, deepar, i_transformer, lag_tst, mqf2, patch_tst, tft, tide and wavenet.

Verified 2026-09-27