Databricks Release Notes

New maintenance updates are available for supported Databricks Runtime versions. These updates include bug fixes, security patches, and performance improvements. For details, see:

• Databricks Runtime 18
• Databricks Runtime 18.2
• Databricks Runtime 18.1
• Databricks Runtime 18.0
• Databricks Runtime 17.3 LTS
• Databricks Runtime 16.4 LTS
• Databricks Runtime 15.4 LTS
• Databricks Runtime 14.3 LTS
• Databricks Runtime 13.3 LTS

You can now govern Databricks-hosted LLMs in Unity Catalog using model services, currently in Beta. A model service is a Unity Catalog securable that represents a governed LLM endpoint, so you can define an endpoint once and share it across workspaces using Unity Catalog privileges, instead of duplicating endpoints per workspace. Databricks provides ready-to-use model services in the system.ai schema for Databricks-hosted foundation models, and you can create your own with the Unity Gateway UI, Catalog Explorer, or the Unity Catalog REST API. See model services.

You can now bring your own Docker container image to AI Runtime CLI (air) workloads. This feature is in Beta. Use a custom image when you need specific system library versions, complex dependencies, or a reproducible environment. See Use custom Docker images with the legacy Python CLI.

Parquet v2 is now generally available for Delta Lake tables in Databricks Runtime 18.1 and above. Parquet v2 uses advanced encodings, v2 data page headers, and INT64 timestamps to improve query performance and reduce table sizes. To use, set the delta.parquet.format.version table property to 2.12.0 or enable automatic upgrades. See Parquet v2 and Automatic upgrades.

The Chat/Agent mode selector has been removed from the Genie Code pane. Genie Code now runs exclusively in Agent mode. If you prefer a conversational response without autonomous task execution (what was previously called Chat mode), include that instruction in your prompt. For example: "Just explain this to me, don't run any code."

See Agent mode.

In a future release, Databricks plans to deprecate classic workspace creation with a Databricks-managed VPC. New classic workspaces will require a customer-managed VPC. For the simplest quick start, use a serverless workspace. A detailed migration timeline will be communicated separately.

See Configure a customer-managed VPC.

The managed Microsoft Dynamics 365 connector in Lakeflow Connect is now generally available. The connector allows you to ingest CRM and ERP data from Dynamics 365 into Databricks using Azure Synapse Link. See Microsoft Dynamics 365 connector.

You can now define parameters in metric views to bind values at query time, serving many query variants from a single definition instead of maintaining a separate metric view per variant. See Use parameters with metric views.

ai_parse_document improves extraction from checkboxes, forms, and complex tables. See ai_parse_document function.

Solution Accelerators are no longer a separate listing type in Databricks Marketplace. The content is unchanged, but they are now categorized as apps (interactive applications) or notebooks (code-based accelerators). See What is Databricks Marketplace?.

The Databricks connector for Microsoft Power Platform is now generally available. See Databricks and Microsoft Power Platform integration.

You can now use ALTER STREAMING TABLE and ALTER MATERIALIZED VIEW with SET OWNER TO to change the owner of streaming tables and materialized views backed by standalone pipelines created in Databricks SQL. See Use ALTER statements with pipeline datasets.

You can now add data quality expectations to materialized views created in Databricks SQL using the CONSTRAINT expectation_name EXPECT (expectation_expr) clause. Expectations let you validate records as they flow through the standalone pipeline that backs the materialized view, and warn, drop, or fail on records that violate a constraint. Streaming tables created in Databricks SQL already support expectations. See Manage data quality with pipeline expectations.

Data Classification is now available by default for workspaces with the compliance security profile enabled, and either HIPAA, HITRUST, PCI-DSS, or FedRAMP Moderate controls or no specific compliance standard selected.

You can now block internet destinations from serverless workloads through the network policies REST API. Blocked destinations are enforced regardless of the policy's network access mode. See Block internet destinations.

Managed agent memory gives your AI agents long-term memory across conversations, with Databricks running the infrastructure and isolating each scope's memories for you. Memory is stored in Unity Catalog memory stores, governed as Unity Catalog securables and accessed through the Unity Catalog REST API. See Managed agent memory.

The Square connector in Lakeflow Connect is now available in Beta. The connector ingests point-of-sale data such as payments, orders, customers, catalog items, and inventory from Square into Databricks. See Square connector.

The Wiz Audit Logs connector in Lakeflow Connect is now available in Beta. The connector allows you to ingest audit log entries, issues, and vulnerability findings from Wiz into Databricks. See Wiz Audit Logs connector.

The Genie Code system table (system.access.assistant_events) now records only user-submitted Genie Code interactions, such as side-panel chat, agent mode, and the inline assistant (%assistant). Background and automated requests, such as autocomplete (ghost text), quick-fix suggestions, and automatic renaming, are no longer logged as events.

This change took effect on May 7, 2026 and corrects earlier behavior that counted these background requests. Event counts recorded before this date can appear higher than actual user activity, so a decrease in daily events around this time is expected and does not indicate a drop in usage.