Product maturity

Status as of October 2026. Production is generally available and covered by support. Preview is in use by customers today, and its interface or behavior can still change before general availability.

Capability	Status
Deployment: AWS, Google Cloud, Microsoft Azure, and bring your own Kubernetes	Production
Ingestion: OneFlow, streaming and database CDC	Production
AI training data pipelines	Production
LakeLog	Production
Onehouse Compute Runtime	Production
Spark jobs on Quanton	Production
SQL clusters and dbt	Production
Table Optimizer	Production
Catalog interoperability: OneSync	Production
Governance catalog, with table and row-level policies	Production
Governance catalog: column policies and column masking	Preview
Lakegres: interactive agentic queries	Production
Notebooks	Production
Open Engines: Trino, Flink, Ray	Preview
AI Gateway	Preview
Semantic layer for natural language queries	Preview
Unstructured data storage	Preview
Vector and text search	Preview

What Preview means for each row:

• Open Engines is launched and open for use.
• AI Gateway ships routing, uptime optimization across providers, and tool-call permissions.
• Unstructured data storage ships on Parquet and Lance.

Availability by cloud

Most of the platform depends only on Kubernetes and object storage and runs the same way everywhere: the data plane, Flows and transformations, table services on every format, SQL, Spark, Notebook, and Lakegres clusters, open engines, the AI capabilities, telemetry export, identity, audit, and every API.

Database CDC runs on any Kafka-compatible streaming endpoint: Apache Kafka, Confluent, Amazon MSK, or LakeLog, the platform-native option, which stages the change log on your object storage without Kafka. The choice is the same for every database source and every cloud.

The cloud-specific items are:

• External catalogs. AWS Glue and the Glue Iceberg REST Catalog are AWS-only. Dataproc Metastore and BigQuery / BigLake are Google Cloud-only. OneLake is Azure-only. Hive Metastore, DataHub, XTable, Databricks Unity, and the Snowflake catalogs work everywhere.
• Object storage file sources. S3 and GCS sources use object event notifications to detect new files. S3-compatible stores that emit bucket notifications, as most enterprise object stores do, work the same way. Where a store does not, Onehouse polls the prefix instead.
• Attached block storage for shuffle spill. Provisioned automatically on AWS and Google Cloud. On bring your own Kubernetes, provide equivalent block storage to your nodes.