CatBoost

Health v2 · maintenance only

Adoption not part of the score

9078 stars · 1324 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

CatBoost is an open-source gradient boosting on decision trees library developed by Yandex, offering high-quality classification, regression, and ranking models with native categorical feature support. It provides fast CPU and multi-GPU training with bindings for Python, R, Java, C++, Rust, Node.js, and Spark.

Use cases

• train a gradient boosting model for tabular data classification
• handle categorical features without manual preprocessing
• train gradient boosted trees on GPU
• build a ranking model for search or recommendations
• export a trained model to ONNX or CoreML for fast inference
• run distributed gradient boosting training with Spark
• compete in Kaggle tabular machine learning competitions

When to choose

• your dataset has many categorical features and you want strong default accuracy
• you need fast GPU or multi-GPU training for gradient boosting
• you need low-latency model inference in C++ or other languages
• you want distributed training via Spark or CLI

When to avoid

• you need deep learning on unstructured data like images or text
• you want a lightweight pure-Python scikit-learn-style tool
• your team is already standardized on XGBoost or LightGBM with no need for categorical handling

Facets

library · maturity stable

machine-learning gpu-computing data-science machine-learning data-science artificial-intelligence python cpp cross-platform cli jvm gradient-boosting decision-trees gbdt categorical-features cuda ranking classification regression spark model-export gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="catboost/catboost")