Tier 5 Fully open / reproducible
What the publisher claims: open source
The one-liner: Automated model selection, fully open.
The six questions
Source and runtimes are downloadable.
Apache 2.0 for the open-source engine.
Runs locally, in Spark or on Kubernetes, and MOJO is a portable artifact.
Search procedure, presets and internals are public and the training data is the user's own.
A skilled group can reproduce AutoML behaviour from the public code and documentation.
H2O-3 has a decade of public roadmap issues, a community forum, and steady merges of outside contributions (18 non-bot external merges in the most recent 67), but H2O.ai still owns the release train and gates the important features in the closed H2O-3 Secure product.
License
Apache 2.0 for open-source H2O-3; H2O-3 Secure (AutoDoc, SSO and RBAC, MOJO scoring into Hadoop and Spark, SLAs) is a closed commercial offering OSI-approved — license source ↗
- Several enterprise features are closed in H2O-3 Secure
What is actually released
- library and distributed runtime source
- AutoML search implementation
- documentation
- MOJO portable artifacts
Why it matters
AutoML is the machine-learning equivalent of a published recipe, and the enterprise tier shows how openness gets carved back into features.
Sources
Derived from the openness dataset behind But Is It Open? — snapshot 2026-11-07 ↗. Scores are qualitative readings of the cited evidence, not verdicts; tier measures openness, not intelligence.