Perpetual
Perpetual MLPerpetual is a gradient boosting machine with a Rust core and Python and R bindings that replaces hyperparameter tuning with a single budget parameter: raising the budget fits a stronger model in one run. It covers classification, regression and ranking, and builds in causal-effect estimation, continual learning, calibrated prediction intervals and drift monitoring, with export to XGBoost and ONNX formats.
Openness
5 high confidence- license
- Apache-2.0 (OSI)
- source
- public (github.com/perpetual-ml/perpetual)
- core features withheld
- no — no paid edition; the Rust core and both language packages are in the repository
Apache-2.0 and built from the public repository, which holds the Rust core and the Python and R packages. The documentation describes no paid edition, and nothing in the repository is held back.
- https://perpetual-ml.github.io/perpetual/ recorded 2026-11-12
Documentation home: Python, R and Rust packages, API reference, causal ML, calibration, drift detection and continual learning guides; no pricing, enterprise or hosted offering.
- https://pypi.org/pypi/perpetual/json recorded 2026-11-12
PyPI record for perpetual 2.5.0, uploaded 2026-11-11, license_expression "Apache-2.0", Repository github.com/perpetual-ml/perpetual.
- https://raw.githubusercontent.com/perpetual-ml/perpetual/main/LICENSE recorded 2026-11-12
LICENSE is the Apache License, Version 2.0, January 2004.
- https://ungh.cc/repos/perpetual-ml/perpetual/files/main recorded 2026-11-12
Repository tree of 308 paths: src/ (Rust core), package-python/, package-r/, benches/ and examples/; no ee, enterprise, commercial or proprietary directory.
Adoption
2 high confidenceMeasured on monthly PyPI downloads of the perpetual package. The R and Rust packages are not counted.
- https://pypistats.org/api/packages/perpetual/recent recorded 2026-11-12
last_month 12,537 downloads of perpetual
Capability
2 high confidencePerpetual is one learner family, gradient-boosted trees, as XGBoost is. Its budget setting removes the tuning loop for that learner rather than searching or ensembling across different models.
- https://perpetual-ml.github.io/perpetual/ recorded 2026-11-12
Documentation: "PerpetualBooster automatically controls step size and node-split generalization via a single intuitive budget parameter"; objectives for classification, regression and learning-to-rank.
- https://raw.githubusercontent.com/perpetual-ml/perpetual/main/README.md recorded 2026-11-12
README: "PerpetualBooster is a gradient boosting machine (GBM) that doesn't need hyperparameter optimization unlike other GBMs"; features list classification, regression and ranking, causal ML, drift monitoring, continual learning, native calibration, and export to XGBoost or ONNX.
Verified 2026-11-12