Darts
Unit8Time-series forecasting and anomaly-detection library from Unit8 with one fit/predict interface over many model kinds - baselines, statistical models such as ARIMA, ETS, Theta and TBATS, regression models on scikit-learn, LightGBM, XGBoost and CatBoost, deep models such as N-BEATS, N-HiTS, TFT, TiDE and TSMixer, and pretrained foundation models such as Chronos-2 and TimesFM that forecast without training. It adds backtesting, ensembling, conformal prediction and covariate support.
Openness
5 high confidence- license
- Apache-2.0(OSI)
- source
- public(github.com/unit8co/darts)
- core features withheld
- no — Unit8 sells consulting and support engagements, not an edition
Apache-2.0 and built from the public repository. Unit8, which built and maintains it, sells consulting projects and support retainers around Darts, but no feature of the library is held back for paying customers.
- https://pypi.org/pypi/darts/json recorded 2026-09-27
PyPI record for darts 0.47.0, uploaded 2026-09-04, license_expression Apache-2.0, Repository github.com/unit8co/darts.
- https://raw.githubusercontent.com/unit8co/darts/HEAD/LICENSE recorded 2026-09-27
LICENSE is the Apache License, Version 2.0, "Copyright 2019 Unit8 SA".
- https://ungh.cc/repos/unit8co/darts/files/master recorded 2026-09-27
Repository tree of 483 paths with no ee, enterprise, commercial or pro directory.
- https://unit8.com/darts-open-source/ recorded 2026-09-27
Unit8 page: "Unit8 built Darts and actively maintains it. Our structured engagements help you unlock its full potential"; engagements of 4-5, 5-6 and 9-12 weeks and a "Support & Expert Access" retainer "Available to any team using Darts".
- https://unit8.com/darts/ recorded 2026-09-27
Unit8 Darts page: "Darts is an open-source Python solution for time series forecasting and anomaly detection" and "Entirely open-source".
Adoption
3 high confidenceMeasured on monthly PyPI downloads of the darts package. Unit8 also publishes the same library as u8darts, with a smaller count that is not added here.
- https://pypistats.org/api/packages/darts/recent recorded 2026-09-27
last_month 153,864 downloads of darts
- https://pypistats.org/api/packages/u8darts/recent recorded 2026-09-27
last_month 56,994 downloads of u8darts
Capability
4 medium confidenceDarts puts dozens of forecasting models behind one interface, and its foundation models forecast a series the user brings with no training run, as TabPFN predicts on a new table. Choosing among its other models, and tuning them, is still the user's work; it does not search for a model automatically.
- https://raw.githubusercontent.com/unit8co/darts/HEAD/README.md recorded 2026-09-27
README: "Darts is a Python library for user-friendly forecasting and anomaly detection on time series"; model table with Baseline, Statistical, Regression, Deep Learning, Ensemble and "Foundation Models ... No training required" (Chronos2Model, TimesFM2p5Model, TimesFM3Model, TiRexModel) sections.
Verified 2026-09-27