GPU installation

CTBoost reports CUDA support at runtime:

import ctboost
print(ctboost.build_info())

Train on CUDA by setting task_type="GPU":

from ctboost import CTBoostClassifier

model = CTBoostClassifier(task_type="GPU", iterations=500)
model.fit(X_train, y_train)

CTBoost 0.1.61 provides unified Linux x86-64 and Windows AMD64 wheels for CPython 3.10 through 3.14. The ordinary installation supports CPU training and CUDA training when a compatible NVIDIA device and driver are available.

python -m pip install -U ctboost

Released artifacts are authoritative

Check the 0.1.61 release for the available Python and platform builds. macOS and ARM wheels are CPU-only. A CUDA-enabled build still works for CPU training on a machine without an NVIDIA GPU. The bundled CUDA runtime remains subject to the NVIDIA CUDA Toolkit license included in each CUDA-enabled wheel.

The unified wheels bundle the CUDA 12.8 runtime; a local CUDA toolkit is not required. GPU training needs a compatible NVIDIA driver and compute capability 6.0 or newer. Linux aarch64, macOS, and CPython 3.8/3.9 artifacts are CPU-only.