Models on records
Gradient boosting on tables such as inspection and fleet data.
We have no published LightGBM build. Our Python 3.13 back ends hold the records such a model learns from, and we say plainly that none of them trains one.
Gradient boosting on tables such as inspection and fleet data.
A FastAPI service in front of a trained model.
A prediction that shows what drove it.
Moves onto 4.7.
Illustration. A dashed outline marks where the technology would sit. We have no published build on it.
A minor release.
The next minor release.
Marked latest. The notes say the project moved GitHub orgs from Microsoft to lightgbm-org.
Source: LightGBM releases on GitHub, checked 6 November 2026. The list shows day and month, and the asset timestamps give 2026 for 4.7.0.
Stops, legs and telematics in PostGIS and Redis, the data a boosted model would learn from.

A graded brief with six scored axes, where every score carries its evidence.
XGBoost
CatBoost
scikit-learn
Python
Apache Airflow
DockerLightGBM 4.7.0, marked latest on GitHub and released on 18 July 2026. Version 4.6.0 came out on 15 February.
No. We say so on the first call and show the nearest work, which is Python data builds with PostGIS and a sync queue.
Yes. The 4.7.0 release notes say the project changed GitHub orgs from Microsoft to lightgbm-org.
You do, from the first commit. Repositories and cloud accounts stay in your name.
Thirty minutes on your data, your labels and your LightGBM version. No sales pitch.