Skip to content
SPECIFICTION
Get listed

AutoKeras

autokeras.com

Free plan Open source WindowsmacOSLinuxAPISelf-hosted
Visit autokeras.com
Spec score 8.9 #2 of 30 in AutoML Software

What it is

AutoKeras is a free, open-source AutoML system based on Keras, designed to make machine learning accessible through code. It supports image and text classification and regression, as well as classification and regression for structured data. Its documentation also includes a tutorial for multi-task or multimodal datasets. Users can assemble custom models from AutoKeras blocks and let the system search for detailed configurations. Image workflows can set architecture type, normalization and data augmentation, or leave those choices for automatic tuning. The best model found can be exported as a Keras Model. The Python package is installed with pip and supports Linux, macOS and Windows, including self-hosted use. The installation guide supports Python 3 and recommends PyTorch version 2.3.0 or later with the PyTorch backend. By default, the final 20% of training data is used for validation, though users can set a different split or supply validation data. Model selection uses validation loss by default, with support for customized metrics and objectives. The project directs users to GitHub Discussions for questions.

Spec sheet

What it does
AutoKeras is an AutoML library that automates deep-learning model selection and hyperparameter tuning.Read 15 Oct 2026 · autokeras.com
Purpose
AutoKeras is an AutoML system based on Keras, developed to make machine learning accessible to everyone.Read 4 Oct 2026 · autokeras.com
Security
The official pages opened for this research do not state security certifications or compliance claims.Read 4 Oct 2026 · autokeras.com
Support
The project directs users to GitHub Discussions to ask questions.Read 4 Oct 2026 · autokeras.com
Maker
The package is developed by DATA Lab at Texas A&M University with Keras-team collaboration.Read 15 Oct 2026 · autokeras.com
Community funding
Open Collective reports $2,065.59 raised and $205.67 disbursed for AutoKeras.Read 15 Oct 2026 · opencollective.com
Community size
Open Collective lists 32 contributors.Read 15 Oct 2026 · opencollective.com
Company location
The Texas A&M Engineering page identifies the institution in College Station, Texas.Read 15 Oct 2026 · engineering.tamu.edu
Custom search spaces
Users can connect built-in blocks or implement their own Block subclasses to customize the search space.Read 15 Oct 2026 · autokeras.com
Deployment ecosystem
The paper states exported models can be deployed through TensorFlow ecosystem tooling.Read 15 Oct 2026 · jmlr.org
Documentation depth
The documentation includes task tutorials, API reference pages, customization guidance, export instructions and an FAQ.Read 15 Oct 2026 · autokeras.com
Funding
Texas A&M reported that the project had been funded by DARPA through an AFRL contract.Read 15 Oct 2026 · engineering.tamu.edu
Image classification
The task API supports image-classification workflows.Read 15 Oct 2026 · autokeras.com
Image features
Image workflows can configure architecture type, normalization, and data augmentation, or leave choices to be tuned automatically.Read 4 Oct 2026 · autokeras.com
Image regression
The task API supports image-regression workflows.Read 15 Oct 2026 · autokeras.com
Installation
The official installation method is pip install autokeras, with Python 3 required.Read 15 Oct 2026 · autokeras.com
Latest package release
PyPI lists version 3.0.0 as released on November 12, 2025.Read 15 Oct 2026 · pypi.org
Licence
The source repository and package metadata specify Apache License 2.0.Read 15 Oct 2026 · github.com
Limit: structured-data production fit
The same Reddit discussion reported relying on boosted trees for tabular data and not finding neural-network success in production.Read 15 Oct 2026 · reddit.com
Maintainer
The package is developed by DATA Lab at Texas A&M University in collaboration with keras-team for version 1.0 and above.Read 4 Oct 2026 · autokeras.com
Model design
Users can build custom models from AutoKeras blocks while letting the system search for detailed configurations.Read 4 Oct 2026 · autokeras.com
Model export
The best searched model can be exported as a Keras model.Read 15 Oct 2026 · autokeras.com
Model selection
AutoKeras uses validation loss by default to select the best model, and supports customized metrics and objectives.Read 4 Oct 2026 · autokeras.com
Multi-modal learning
AutoModel can combine different input types such as images and text.Read 15 Oct 2026 · autokeras.com
Multi-task learning
AutoModel can produce multiple prediction heads, including classification and regression outputs.Read 15 Oct 2026 · autokeras.com
Multimodal
The documentation includes a tutorial for working with multi-task or multi-modal datasets.Read 4 Oct 2026 · autokeras.com
Notable news: 2019
Texas A&M announced AutoKeras as an open-source machine-learning breakthrough on January 22, 2019.Read 15 Oct 2026 · engineering.tamu.edu
Praise: accessible experimentation
A Reddit user reported using AutoKeras during early experimentation because cloud AutoML services felt too complicated.Read 15 Oct 2026 · reddit.com
Preprocessing search
The documented search space can tune preprocessing such as normalization and image augmentation.Read 15 Oct 2026 · jmlr.org
Public API
The AutoModel API exposes fit, predict, evaluate and export_model methods.Read 15 Oct 2026 · autokeras.com
Rating on Microsoft Marketplace
AutoKeras is shown at 5.0 out of 5 from 1 rating.Read 15 Oct 2026 · marketplace.microsoft.com
Requirements
The installation guide says AutoKeras supports Python 3 and recommends PyTorch version 2.3.0 or later with the PyTorch backend.Read 4 Oct 2026 · autokeras.com
Resume interrupted runs
Runs can resume automatically when overwrite remains false and the same code is rerun.Read 15 Oct 2026 · autokeras.com
Resume runs
With overwrite set to false, running the same code again continues the previous fit from the directory contents.Read 4 Oct 2026 · autokeras.com
Scaling limitation
The JMLR paper identifies improved distributed search and training for larger datasets as future work.Read 15 Oct 2026 · jmlr.org
Scope limitation
The JMLR paper says AutoKeras focuses on deep neural-network models and raw data types rather than tree-based model families.Read 15 Oct 2026 · jmlr.org
SDK and dependencies
The package depends on Keras, Keras-Tuner, packaging and dm-tree.Read 15 Oct 2026 · github.com
Structured-data classification
The task API supports structured-data classification.Read 15 Oct 2026 · autokeras.com
Structured-data regression
The task API supports structured-data regression.Read 15 Oct 2026 · autokeras.com
Support channel
The project directs users to GitHub Discussions for questions.Read 15 Oct 2026 · autokeras.com
System requirements
The installation page recommends PyTorch 2.3.0 or newer and the current package metadata requires Python 3.8 or newer.Read 15 Oct 2026 · autokeras.com
Tasks
Its supported tasks include image and text classification and regression, plus structured data classification and regression.Read 4 Oct 2026 · autokeras.com
Text classification
The task API supports text-classification workflows.Read 15 Oct 2026 · autokeras.com
Text regression
The task API supports text-regression workflows.Read 15 Oct 2026 · autokeras.com
Training behavior
By default, the last 20% of training data is used for validation, and users can provide a different split or their own validation data.Read 4 Oct 2026 · autokeras.com
Tuning algorithms
AutoModel documents greedy, Bayesian, Hyperband and random tuner choices.Read 15 Oct 2026 · autokeras.com

Plans and prices

AutoKeras Free Python package installed with pip Read 4 Oct 2026 · autokeras.com
Open-source library Free Install locally with pip · no hosted-service limits stated Read 15 Oct 2026 · opencollective.com

Every AutoKeras plan and price →

Strong points and limits

Strong points

  • Free Python package installed with pip
  • Supports image, text and structured-data tasks
  • Can export the best model as a Keras Model
  • Users can customize validation and selection objectives

Limits

  • Workflow is code-based
  • Installation guide recommends PyTorch 2.3.0 or later for its PyTorch backend
  • Default validation uses the final 20% of training data

Who it suits

AutoKeras suits people building machine-learning models through code who want automated model selection and configuration search. It supports several image, text and structured-data tasks.

Verdict

AutoKeras provides automated model search for several common data tasks while allowing custom blocks and selection objectives. It is a code-based Python package, so it suits users comfortable working with that workflow.

Where Specifiction places it

Alternatives to AutoKerasAll alternatives →