Databricks Release Notes

The workspace sidebar navigation has updated section groupings:

• A new Analytics section groups Dashboards, Genie Agents, and Alerts.
• The SQL section is renamed to Lakehouse and contains SQL Editor, Queries, Query History, and SQL Warehouses.

Workspace sidebar before and after the navigation updates.

In the app switcher, Lakehouse is also renamed to Analytics and AI.

The app switcher showing the Lakehouse option updated to Analytics and AI.

The Databricks CLI is now generally available (GA). As part of this release, the CLI has been updated to store user-to-machine (U2M) tokens in OS-native secure storage. See Token storage.

For information about the Databricks CLI, see Databricks CLI.

The system.compute.instance_events and system.compute.instance_pools system tables are now in Public Preview. Use these tables to track state transitions of classic compute instances and the full history of instance pool configurations across your account. See Instance events table schema and Instance pools table schema.

New Lakebase projects now have native Postgres password authentication disabled by default. Existing projects are unaffected. To allow native Postgres roles to connect with static passwords, enable password connections in project settings. See Manage password connections.

Vector Search storage-optimized endpoints are now generally available. Storage-optimized endpoints have a larger capacity (over one billion vectors at dimension 768) and provide 10-20x faster indexing than standard endpoints. Pricing is optimized for the larger number of vectors. For details, see Vector Search endpoint options.

Fixed level of detail (LOD) expressions in metric views are now defined directly in the expr field of a dimension using SQL window functions, replacing the previous two-step approach of pre-computing values in the source query. See Use level of detail (LOD) expressions in metric views.

You can now serve custom and fine-tuned LLMs on Databricks with a vLLM engine, including multimodal models and PEFT recipes that Foundation Model APIs does not support. Custom LLM Serving is in Beta. See Serve custom LLMs with Custom Model Serving.

You can now configure Microsoft Entra ID as the identity provider for single sign-on (SSO) to Databricks from Power BI using an account-level federation policy. See Configure Microsoft Entra ID SSO for Power BI.

You can now use the OAuth U2M flow to obtain tokens as a service principal's identity instead of your own user account. You must have the Service Principal Manager role on the service principal.

See Authenticate as a service principal using OAuth U2M.

Customer-managed keys (CMK) for Lakebase are now available in all Lakebase regions on AWS and Azure. CMK encrypts Lakebase project data at rest using a key your organization owns in its cloud KMS. CMK is available for new Lakebase projects. See Customer-managed keys for Lakebase.

Lakebase now supports compute sizes up to 64 CU (128 GB RAM), up from the previous maximum of 32 CU (64 GB RAM). Larger workloads can now use dynamic scaling. See Autoscaling.

You can now use the following MLflow managed evaluation features in workspaces encrypted with customer-managed keys (CMK):

• Production monitoring of AI apps with MLflow 3 scorers.
• The Review App for collecting expert feedback on traces.
• Evaluation datasets. The catalog that stores the dataset must not itself be CMK-encrypted.

You can now add another Supervisor Agent as a subagent tool, enabling nested multi-agent orchestration. See Use Supervisor Agent to create a coordinated multi-agent system.

Databricks Container Services for standard compute is now in Beta. You can now specify a Docker image when creating standard compute on Databricks Runtime 18.3 or later, allowing you to use custom containers in shared compute environments. See Databricks Container Services for standard compute.

Model Serving now supports Google Gemini 3.5 Flash as a Databricks-hosted model. You can access this model using Foundation Model APIs.

Customers are responsible for ensuring their compliance with applicable terms.

Edits to jobs and pipeline settings in the UI now automatically update YAML. See Edit bundle resources.

The Databricks Add-in for Excel (Public Preview) now supports all compliance security profile standards supported by Databricks. For more, see Compliance security profile and Connect to Databricks from Microsoft Excel.

You can now view all of your job and pipeline runs in the updated Runs list. Track all of your pipeline runs in one place and filter by status, time, run-as user, run ID, and error codes in real time. Spot trends using the visualization and summary of the current top five error codes. See View recent runs across all jobs and pipelines.