H2O-3
The open-source distributed ML and AutoML engine that 20,000+ organizations and 2M+ data scientists run. Available in two offerings: free and open source under the Apache 2.0 license, or H2O-3 Secure, the same code with enterprise controls, AutoDoc, and support to run it in production.
- H2O-3 Secure is the same code and APIs as open-source H2O-3, plus enterprise packages, MOJO scoring, AutoDoc, SSO/RBAC, controls aligned to the FedRAMP High baseline, CVE patching, and SLAs.
- Open-source H2O-3 is Apache 2.0 and used by 20,000+ organizations and 2M+ data scientists; it does not include AutoDoc.
- AutoDoc is the upgrade lever from open source to Secure: the documentation model risk asks for, generated with the model.
- Nothing you built on H2O-3 changes; Secure adds controls around it.
One engine, two ways to run it.
Choose the offering that meets your needs. One codebase, different ways to deploy.
Machine learning for everyone.
- PythonWorks with pandas, scikit-learn, and Jupyter. Install with pip install h2o.
- RWorks with the R ecosystem you know. Install with install.packages("h2o").
- Java / ScalaRun H2O directly on the JVM and integrate into existing JVM applications.
- H2O FlowA browser-based notebook UI that runs on top of H2O-3. No coding required.
Train your first AutoML model in minutes.
Follow these simple steps to install H2O-3 and train your first model.
- 01Install H2O-3Bash / ShellDownload
pip install h2o - 02Train your first AutoML modelPython
import h2o from h2o.automl import H2OAutoML h2o.init() data = h2o.import_file("your_data.csv") aml = H2OAutoML(max_runtime_secs=60, seed=1) aml.train(y="target_column", training_frame=data) print(aml.leaderboard)
What Secure adds to open source.
H2O-3 open source is for experimentation and research. H2O-3 Secure is the same engine with the packaging an enterprise needs to bet production on it.
- Hardened and patchedControls aligned to the FedRAMP High baseline, CVE patching on a committed cadence, and security bulletins you can hand to your CISO.
- SSO and RBACEnterprise identity, role-based access, and an audit of who ran what. The controls open-source H2O-3 does not ship.
- AutoDocModel documentation for H2O-3 models, generated with the model, in the format a reviewer asks for. The upgrade lever from OSS.
- MOJO everywhereScore H2O-3 models inside Hadoop, Spark, Teradata, and Oracle without moving data, on the same MOJO runtime as H2O Driverless AI, our AutoML.
- Support with SLAsEnterprise support from the engineers who build H2O-3, with response-time commitments and security bulletins.
- Zero code changeDistributed ML and AutoML on Hadoop, Spark, and Kubernetes, in Python, R, Java, and Scala. Every notebook you wrote still runs.
H2O-3 OSS vs H2O-3 Secure
All H2O-3 algorithms
H2O AutoML
Multi-node distributed model training
Model scoring
PyPi and R-CRAN packages
Community support
Early access to latest algorithms and improvements
Hadoop and Kubernetes enterprise packages
Production MOJO scoring for Hadoop, Spark, Teradata, Oracle
Managed distributed XGBoost service
Model reproducibility
AutoDoc automated model documentation
Enterprise SSO and secrets management
RBAC and audit capabilities
Controls aligned to the FedRAMP High baseline
Premium support with SLAs
Commercial CVE patching
Controls aligned to the FedRAMP High baseline. Alignment is not certification; the FedRAMP Certified offering is H2O AI Cloud for Government.
From notebook to production without a rewrite.
- 01Keep the codeYour H2O-3 scripts, AutoML runs, and Sparkling Water jobs run unchanged, exactly as they do today.
- 02Add the controlsSwitch on SSO, RBAC, and the hardened packages, and enrol in CVE patching on a committed cadence.
- 03Document and scoreAutoDoc for the reviewer. MOJO into the database or Spark cluster where the data already lives.
- 04Run it with supportSLAs, security bulletins, and H2O AI support engineers on call from the team that builds H2O-3.
A training engine in the predictive lane, and the open-source root of the platform.
H2O-3 sits beside Driverless AI as a training engine, and the Data Science Agent runs it in Notebook Labs. Highlighted is H2O-3.
What ships today
You operate it, or H2O AI operates it for you.
Same platform either way: the same capabilities, the same governance, and the same audit trail. The choice stays open, so you can switch between them as the business changes, without rebuilding what runs on top.
H2O AI Cloud for Government
The H2O AI Cloud for Government has achieved FedRAMP High Certification, delivering the highest standard of cloud security so federal agencies can harness predictive and generative AI without compromising data sovereignty or compliance.
H2O-3 in production
:quality(75))
:quality(75))
:quality(75))
Questions people ask before they call us.
H2O-3 is the open-source distributed ML and AutoML engine under Apache 2.0. H2O-3 Secure is the same code with enterprise packages, MOJO scoring into Hadoop, Spark, Teradata, and Oracle, AutoDoc, SSO and RBAC, controls aligned to the FedRAMP High baseline, CVE patching, and support with SLAs.
Get the production controls without touching your H2O-3 code.
We will map your current open-source footprint to H2O-3 Secure, show the security and audit deltas, and price the subscription.