GitHub Repository

Ollama for classical ML models. AOT compiler that turns XGBoost, LightGBM, scikit-learn, CatBoost & ONNX models into native C99 inference code. One command to load, one command to serve. 336x faster than Python inference.

683 starsPython

Timber – Ollama for classical ML models, 336x faster than Python

Name: Timber – Ollama for classical ML models, 336x faster than Python
Availability: InStock
Author: kossisoroyce

by kossisoroyce·Mar 2, 2026·207 points·33 comments

Visit Project View on HN

AI Analysis

●●●BangerWizardrySolve My Problem

336× faster tree model inference; compiles sklearn/XGBoost to C99, serves like Ollama.

Strengths

•338x latency improvement over Python with microsecond-scale native calls and no runtime overhead
•AOT compilation to portable C99 removes Python dependency entirely, enabling edge/IoT/regulated deployments
•Rigorous benchmark methodology with reproducible scripts; transparent comparison table vs ONNX Runtime/Treelite/lleaves

Weaknesses

•Only accelerates tree-based models; deep learning and non-tree ensembles need alternative solutions
•Very early (0 stars/forks); adoption and long-term maintenance unproven