AutoGluon

AutoGluon is an open-source Python library that automates model building across tabular data, time series, and multimodal tasks. It suits Python users who want automated modeling without a software fee, especially those working with predictive or forecasting workflows. Its range is a strength; macOS users who need GPU acceleration should look elsewhere.

Overview

AutoGluon aims to reduce the work between raw data and predictive models. For tabular classification and regression, it handles data cleaning, feature engineering, hyperparameter optimization, and model selection. That makes it a practical option for users who want a strong automated baseline without assembling each stage themselves, though the workflow remains centered on Python.

Beyond tables, it supports forecasting across multiple time series and multimodal prediction tasks. The breadth is useful when projects span different data types, but AutoGluon is a library rather than a ready-made desktop application.

Key features

Automated tabular modeling

AutoGluon trains classification and regression models while automating several demanding parts of the pipeline, from data preparation through model selection. That can save implementation effort, while users who need full control over each modeling choice may prefer a more manual workflow.

Time-series forecasts

TimeSeriesPredictor trains multiple models to produce probabilistic forecasts for multiple series, using historical observations and related covariates. Its model mix includes ETS and ARIMA from StatsForecast, LightGBM, DeepAR and Temporal Fusion Transformer from GluonTS, and Chronos. This breadth suits forecasting work where uncertainty and multiple related series matter.

Multimodal tasks

AutoMM reduces manual preprocessing, model selection, and fine-tuning when adapting foundation models to domain-specific data. Examples cover image and text prediction, object detection, image segmentation, document prediction, entity extraction, and semantic matching. It can adapt models from Hugging Face, TIMM, and MMDetection, making it relevant to users who need more than conventional tabular modeling.

Pricing

AutoGluon costs nothing. The library is open source under the Apache 2.0 license, with no trial required. This is a strong fit for individuals and teams comfortable installing and operating a Python library; there is no paid tier to unlock in the listed plan.

Cloud use can still incur costs: AutoGluon-Cloud is described as a free wrapper, but SageMaker compute and S3 storage are billed to the user's AWS account. AutoGluon is pre-installed in all releases of Amazon SageMaker Distribution, and SageMaker Autopilot is described as a managed AutoGluon experience.

Platforms

AutoGluon supports Linux, macOS, and Windows and requires Python 3.10 through 3.13. It can be self-hosted, and the project points to AutoGluon Cloud and deep learning containers for cloud training and deployment. GPU use is not supported on macOS, so users who need GPU acceleration must use Linux or Windows.

Who it's for

Choose AutoGluon if you work in Python and want automated workflows for tabular prediction, time-series forecasting, or multimodal tasks without paying for the library. It is particularly well matched to users who value broad modeling coverage and can manage their own development environment. It is less suitable for readers seeking a graphical, non-coding product or macOS GPU support.

Pros and cons

• Pro: One library covers tabular, time-series, and multimodal workflows, reducing the need to switch tools across these task types.
• Pro: Automated cleaning, feature engineering, tuning, and model selection reduce pipeline work for tabular prediction.
• Pro: The open-source Apache 2.0 library has a free plan.
• Con: Python 3.10 through 3.13 is required, so it is not a plug-and-play choice for non-coders.
• Con: macOS GPU use is unsupported, ruling it out for users who require local GPU acceleration on that platform.
• Con: Cloud compute and storage through SageMaker still generate AWS charges despite the free wrapper.

Alternatives

Browse AutoML Software or Predictive Analytics Software for more options in those categories.

• BigML is worth considering for a web- and API-accessible option with a free monthly plan; its FREE plan limits datasets to 16 MB per task, allows two parallel tasks, and includes one user.
• FEDOT is another free open-source AutoML framework, with Linux, macOS, Windows, and self-hosted support.
• JADBio offers a web and API route with a free Basic plan capped at one seat, three projects, 50 MB uploads, 500 MB storage, and one model export.
• H2O Driverless AI is a commercial-license option for readers willing to contact sales rather than use a free plan.
• FLAML is another free option for users seeking an AutoML tool across Linux, macOS, Windows, API, or self-hosted environments.
• LightAutoML is a free, Apache 2.0 Python library installable from PyPI, making it a comparable open-source route.
• Auto-PyTorch is a free alternative for Linux or self-hosted environments.
• DataRobot may suit teams wanting a self-service SaaS trial, with up to five team members and an account-specific number of shared modeling workers for 30 days.

Verdict

AutoGluon is a compelling free choice for Python users who want automated modeling across tables, time series, and multimodal data. Its broad workflow coverage and open-source license are the main reasons to choose it. Look elsewhere if you need macOS GPU support, a non-coding interface, or cloud usage without separate AWS compute and storage charges.