FLAML
A fast Python library for automated machine learning and hyperparameter tuning.
Automate machine learning workflows and hyperparameter optimization under resource constraints.
FLAML is a lightweight Python library designed to automate machine learning model selection and hyperparameter tuning efficiently. Data scientists and ML engineers looking to optimize performance under strict computational limits will find it particularly useful.
What it is
The library automates workflow generation for classification, regression, and generic hyperparameter tuning tasks using custom or popular learners like XGBoost and LightGBM. It is powered by research from Microsoft Research and collaborators, handling large search spaces with heterogeneous evaluation costs and early stopping.
Who it's for
It suits data scientists and machine learning practitioners seeking rapid model prototyping and efficient resource utilization. It does not suit those looking for the AutoGen multi-agent framework, which has been moved to a separate repository.
Key features
Task-Oriented AutoML
Finds quality models for classification and regression tasks using a scikit-learn style estimator.
Hyperparameter Tuning
Tunes hyperparameters for specific learners like XGBoost, LightGBM, and Random Forest or custom functions.
Zero-Shot AutoML
Sets automatic high-performance hyperparameter configurations while using standard training APIs.
Microsoft Fabric Support
Provides built-in AutoML and hyperparameter tuning within Microsoft Fabric Data Science.
Up and running in 3 steps
Our honest verdict
What's great
- Significantly speeds up model selection and hyperparameter optimization under resource constraints.
- Provides a scikit-learn compatible interface with minimal code required for quick start.
- Supports customized learners and generic user-defined function tuning.
- Integrates smoothly with MLflow and Microsoft Fabric data science environments.
Where it falls short
- Requires Python version between 3.10 and 3.13 for full model support compatibility.
- Specific AutoML features require installing extra pip dependencies beyond the minimal setup.
- AutoGen features have been completely removed from this repository and require a separate install.
Pricing & license
FLAML is completely free under its open-source license. The only cost is running the computations on your own hardware or cloud infrastructure.
Open remote AI-training roles
874 open remote AI-training roles, updated November 15, 2026; most pay $20–$170 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
These reviewed AI tools do the same job instantly — from our scored directory.
Use it, or have us set it up
FAQ
Does FLAML still include AutoGen?
No, AutoGen has moved to a dedicated GitHub repository and is no longer part of the FLAML codebase.
What Python versions are officially required?
The latest version of FLAML requires Python between 3.10 and 3.13.
Can I use FLAML for custom evaluation functions?
Yes, you can run generic hyperparameter tuning for any custom function using flaml.tune.
