Databricks Release Notes

You can now add other Databricks apps and Unity Catalog tables as resources for your Databricks app. App resources enable app-to-app communication, and table resources provide governed access to structured data. See Add a Databricks app resource to a Databricks app and Add a Unity Catalog table resource to a Databricks app.

You can now group multiple SQL statements across multiple tables into a single atomic transaction using BEGIN ATOMIC ... END; syntax. All changes succeed together or roll back together, ensuring data consistency across your operations. Multi-table transactions require catalog commits enabled on participating tables and are supported on Unity Catalog managed Delta tables.

See Transactions.

Compute sizing for Databricks Apps now supports all compliance security profile standards supported by Databricks and is available by default for workspaces with the compliance security profile enabled.

For a full list of supported compliance standards, see Compliance security profile.

Inbound Private Link for performance-intensive services such as Zerobus Ingest and Lakebase Autoscaling is now in Public Preview. See Configure inbound PrivateLink for performance-intensive services.

Mosaic AI Model Serving now supports Google Gemini 3.1 Flash Lite as a Databricks-hosted model.

To access this model, use:

• Foundation Model APIs pay-per-token
  • Query vision models
• Batch inference workloads using AI Functions

You can now configure endpoint telemetry to persist logs, traces, and metrics from custom model serving endpoints into Unity Catalog Delta tables using OpenTelemetry. This enables root cause analysis, application health monitoring, and compliance retention using standard SQL queries.

See Persist custom model serving data to Unity Catalog.

You can now configure telemetry for your Databricks apps to collect traces, logs, and metrics and persist them to Unity Catalog tables using the OpenTelemetry (OTel) protocol. After you enable app telemetry, Databricks automatically captures system logs and usage events.

See Configure telemetry for Databricks Apps.

Mosaic AI Model Serving now supports OpenAI GPT-5.4 as a Databricks-hosted model. You can access this model using Foundation Model APIs pay-per-token.

Customers are responsible for ensuring their compliance with the terms of OpenAI's Acceptable Use Policy.

The workspace browser side panel now includes a tree view that provides a flexible, tree-based view of your workspace folder hierarchy. The tree view allows you to expand descendant folders, change the root of the tree, and navigate your workspace while keeping the tree steady. This improves the multi-asset authoring experience by providing better control over workspace navigation.

See Tree view of the workspace browser side panel (Public Preview).

Lakebase is now generally available for workspaces with the compliance security profile when the compliance standard is set to HIPAA, C5, TISAX, or None.

See Lakebase compliance.

Supervisor Agent is now available in additional regions. For the full list of supported regions, see Features with limited regional availability.

Knowledge Assistant is now available by default for workspaces with the compliance security profile enabled and HIPAA controls selected.

Use Knowledge Assistant to create a chatbot that can answer questions about your documents and provide high-quality responses with citations.

See Use Knowledge Assistant to create a high-quality chatbot over your documents.

Databricks now provides managed OAuth flows for select external MCP servers, eliminating the need to register your own OAuth app or manage credentials. Supported providers include Glean MCP, GitHub MCP, Google Drive API, and SharePoint API.

See Install an external MCP server.

You can now apply tags to notebooks to organize and categorize them for easier management. Notebooks also support governed tags, including certification and deprecation system tags, to indicate trust levels or lifecycle status directly within the notebook interface. See Apply tags to notebooks.

You can now protect data in Unity Catalog catalogs with your own encryption keys using customer-managed keys (CMK). See Customer-managed keys for Unity Catalog.

You can now schedule automatic data refreshes in the Databricks Connector for Google Sheets on an hourly, daily, or weekly basis. See Schedule data refreshes in Google Sheets.

The new Databricks Excel Add-in is in Public Preview and allows you to connect Microsoft Excel to your workspace, import data from tables or Metric Views, execute SQL queries, and analyze data directly in Excel. See Connect to Databricks from Microsoft Excel.

Lakebase now includes high availability. A high availability endpoint pairs your primary read/write compute with one to three secondary compute instances across availability zones. If the primary becomes unavailable, Lakebase automatically promotes a secondary and your application continues from the last committed transaction, with no connection string changes. You can optionally serve read traffic from the secondary compute instances for read offload. See High availability.

Lakebase is now available in three additional regions:

• ca-central-1 (Canada Central)
• eu-west-2 (Europe – London)
• sa-east-1 (South America – São Paulo)

See Get started with Lakebase.