Databricks Release Notes

For workspaces enabled for customer-managed keys (CMK) for managed services, the system catalog's data encryption settings can be configured to decrypt the statement_text and error_message fields in the query history system table. See Reading encrypted fields.

Using CREATE, ALTER, or the Lakeflow UI to modify ETL and ingestion pipelines (Spark Declarative Pipelines on Lakeflow and Lakeflow Connect) is now GA. You can modify pipelines to apply row filters, column masks, table and column tags, column comments, and (for materialized views only) table comments.

See ALTER STREAMING TABLE and ALTER MATERIALIZED VIEW. For general information about using ALTER with Spark Declarative Pipelines on Lakeflow, see Use ALTER statements with pipeline datasets.

Mosaic AI Model Serving now supports Anthropic Claude Opus 4.6 as a Databricks-hosted model.

To access this model, use:

• Foundation Model APIs pay-per-token
  • Query reasoning models
  • Query vision models

Default SQL warehouse settings are now generally available. Workspace administrators can set a default SQL warehouse that is automatically selected in SQL authoring surfaces, including the SQL editor, AI/BI dashboards, Genie Agents, Alerts, and Catalog Explorer. Individual users can also override the workspace default by setting their own user-level default warehouse. See Set a default SQL warehouse for the workspace and Set a user-level default warehouse.

You can now connect Databricks Assistant in Agent mode to external tools and data sources through the Model Context Protocol (MCP). The Assistant can use any MCP servers that have been added to your workspace and that you have permission to use.

See Connect Genie Code to MCP servers.

You can now directly select Databricks tables from the Catalog Explorer and import data as pivot tables in Google Sheets using the Databricks Connector. See Databricks Connector for Google Sheets.

For workspaces in ap-northeast-1 (Tokyo) and ap-northeast-2 (Seoul), Genie Agents now use models hosted in the same region. Users in these regions no longer require cross-Geo processing to use Genie Agents.

See Availability of Designated Services in each Geo for more information on which features require cross-geo processing.

You can now apply tags to Databricks apps to organize and categorize them. See Apply tags to Databricks apps. Search is not supported using Databricks apps tags.

The Zendesk Support connector allows you to ingest ticket data, help center content, and community forum data from Zendesk Support. See Zendesk Support connector overview.

Databricks Data Quality Monitoring Anomaly Detection is now in Public Preview. The feature is enabled at the schema level and learns from historical data patterns to detect data quality anomalies. The health of all monitored tables is consolidated into a single system table and a new UI. See Anomaly detection.

TRIGGER ON UPDATE is now generally available. Use trigger on update in Databricks SQL when creating a pipeline schedule to refresh the pipeline when a source table changes.

See Schedule refreshes for more information.

The lakeflow system tables jobs, job_tasks, job_run_timeline, and job_task_run_timeline are now generally available (GA). These tables provide insights into all jobs and job runs in your account.

For a reference of these tables, see Jobs system table reference.

You can now view your past workspace tab sessions and restore them. See Organize your work with spaces.

You can now copy images from your local file system and paste them into Markdown cells in Databricks notebooks using ⌘ + V (Mac) or Ctrl + V (Windows). See Display images.

• Full query result downloads: A new API endpoint supports downloading complete result sets up to about 1 GB.
• Fixed chart generation from prompts: Resolved an issue where requesting charts directly in the prompt would not create visualizations.
• Collapsible error messages: Errors now appear in a collapsible modal for improved readability.
• Catalog permission warnings: Users now see an informative banner when they lack USE CATALOG permissions on attached tables, rather than encountering errors at SQL execution time.

Mosaic AI Model Serving now supports OpenAI GPT-5.2 Codex as a Databricks-hosted model. This code-specialized model excels at code generation, refactoring, debugging, and software engineering tasks. 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.

Serverless workspaces are now generally available. Serverless workspaces come pre-configured with serverless compute and default storage, providing a fully-managed, enterprise-ready SaaS experience. See Create a serverless workspace.

Compute sizing for Databricks Apps is now generally available. Choose between Medium (2 vCPUs, 6 GB) and Large (4 vCPUs, 12 GB) compute sizes to match your workload requirements.

See Configure compute resources for a Databricks app.

App compute sizing is coming soon to workspaces with the compliance security profile enabled.