You can now add AI Function tasks to your Databricks jobs to run ai_parse_document, ai_extract, or ai_classify on the files in a volume or the rows of a table, without writing SQL. Each run processes only the inputs that are new or changed since the previous run and writes the results to a Unity Catalog table that later tasks can read. This feature is in Beta. See AI Function tasks for jobs.
You can now run read queries on tables protected by row filters, column masks, and attribute-based access control (ABAC) policies with Lakehouse//RT. See Fine-grained access control.
You can now run a Lakeflow Job when one or more upstream jobs complete, including across workspaces attached to the same Unity Catalog metastore. Configure dependencies on the downstream job. See Trigger jobs when upstream jobs complete.
When you write to Lakebase from Structured Streaming, you can use merge strategies to control how upserts update individual columns. Keep the first or last value, ignore nulls, or select values by an ordering column. Merge strategies require Databricks Runtime 19 and above. See Merge strategies for upserts.
The managed Okta connector is now available in Beta in Databricks Lakeflow Connect. Workspace admins can enable Lakeflow Connect for Okta on the Previews page. Use the connector to ingest identity and access data from Okta into Databricks, including users, groups, group memberships, app integrations, devices, policies, and administrator roles. The connector uses SSWS API token authentication and s…
Simba JDBC Driver (Legacy) 2.8.7 updates jackson-core, jackson-databind, and jackson-datatype-jsr310 to 2.21.8. Download it from the JDBC drivers archive under All JDBC Driver Versions (Simba).
Ingesting into tables backed by default storage with Zerobus Ingest in Lakeflow Connect is now generally available. See Create or identify the target table.
A schema can now hold up to 1,000 Unity Catalog secrets, and a metastore up to 10,000, up from 100 and 1,000. The limits apply to Databricks-managed secrets and secrets backed by AWS Secrets Manager. BROWSE on a catalog now applies to Unity Catalog secrets, so users can discover secrets and their metadata without being able to read the values or reference the secrets in code. See Privileges for Un…
New features and improvements ALTER SHARE ADD/ALTER VIEW now accepts WITH HISTORY and WITHOUT HISTORY; existing history defaults are unchanged. Library and environment updates Updated Python libraries: pytz from 2025.2 to 2026.5 tzdata from 2026.2 to 2026.5
Analyze files in Unity Catalog volumes (Public Preview): A Genie Agent can now analyze unstructured files stored in Unity Catalog volumes, such as PDFs, slide decks, and images, and answer with citations that point to the source page. Workspace admins can control access from the Previews page. See Analyze files in volumes with a Genie Agent.
Beginning November 1, 2026, Unity Gateway budgets include spend from task-specific AI Functions such as ai_parse_document, ai_extract, and ai_classify across batch and real-time execution paths. See Manage budgets for Unity Gateway.
Beginning November 1, 2026, Unity Gateway budgets include spend from task-specific AI Functions such as ai_parse_document, ai_extract, and ai_classify across batch and real-time execution paths. See Manage budgets for Unity Gateway.
With managed warehouse, users can view traces stored in Unity Catalog in the MLflow UI without selecting a SQL warehouse. MLflow runs trace queries on a SQL warehouse that it creates in your workspace when your workspace supports it, and on serverless compute otherwise. Workspace admins can enable Managed Warehouse for MLflow on the Previews page. See Managed warehouse for MLflow traces.
You can now see how current each destination table is in Lakeflow Connect database connector pipelines, in Public Preview. The new Data freshness column in the pipeline's tables list shows the time since the latest source commit that the pipeline ingested for each change data capture (CDC) table. Data freshness is available for SQL Server, PostgreSQL, MySQL, and Oracle sources. You can also query…
DeepSeek V4 Pro (0813) was retired on October 30, 2026. DeepSeek V4.1 Flash (1017) is the recommended replacement. See Deprecated and retired models for migration guidance.
Thinking Machine Labs Inkling was retired on October 30, 2026. GLM 5.3 or Kimi K3 is the recommended replacement. See Deprecated and retired models for migration guidance.
Moonshot AI Kimi K2.7 was retired on October 30, 2026. Kimi K3 is the recommended replacement. See Deprecated and retired models for migration guidance.
Claude Code auto mode is now supported for both Anthropic and non-Anthropic models hosted by Databricks. Auto mode reviews tool calls to reduce routine permission prompts. See Auto mode.
Relationships between Unity Catalog metric views are in Beta. Use ALTER VIEW ... ADD CONSTRAINT ... RELATIONSHIP to record how a source metric view joins to a target metric view, including the cardinality and the join condition, so that AI assistants and BI tools choose the right join path. Relationships require Databricks Runtime 19 and above, or a SQL warehouse. See Relationships between metric…
You can now track usage of Unity Gateway Skills, which are in Beta. Audit logs record skill list, load, and file fetch events under the aiGatewaySkills service. The Skills tab of the usage dashboard shows downloads, loads, and activity by user.
Claude Code auto mode is now supported for both Anthropic and non-Anthropic models hosted by Databricks. Auto mode reviews tool calls to reduce routine permission prompts. See Auto mode.
You can now track usage of Unity Gateway Skills, which are in Beta. Audit logs record skill list, load, and file fetch events under the aiGatewaySkills service. The Skills tab of the usage dashboard shows downloads, loads, and activity by user.
The managed Microsoft Graph Logs connector is now available in Beta in Databricks Lakeflow Connect. Use the connector to ingest security alerts from Microsoft Defender XDR, Microsoft Sentinel, and other Microsoft security products into Databricks. The connector uses OAuth machine-to-machine authentication with a Microsoft Entra ID application and supports incremental ingestion.
You can now apply attribute-based access control (ABAC) row filters and column masks to views, materialized views, and streaming tables shared through OpenSharing. Shared views are in Beta and require the ABAC on views preview. See OpenSharing and ABAC.
Anomaly detection for data quality monitoring is now generally available. Enable it at the schema level to monitor the freshness and completeness of every table in the schema, based on historical data patterns. Results for all monitored tables are stored in the system.data_quality_monitoring.table_results system table, which is also now generally available. See Anomaly detection.
Starting in December 2026, Databricks-provided MCP connectors in Genie One and Genie Code will be enabled by default for workspaces with the compliance security profile enabled. The connectors include Google Drive, Gmail, Google Calendar, Microsoft 365, Atlassian, Slack, and GitHub. They let Genie One and Genie Code connect to third-party services to search your documents and data and take actions…
Native XML file format support is now generally available. You can ingest, query, and parse XML files in batch or streaming workloads with Auto Loader, read_files, and COPY INTO, and use the from_xml and schema_of_xml SQL functions. See Read and write XML files.
Real-time mode in Lakeflow pipelines is now generally available on classic compute and serverless compute. Use it for operational workloads that need end-to-end latency as low as five milliseconds, such as fraud detection. See Use real-time mode in Lakeflow pipelines.
Simba JDBC Driver (Legacy) 2.8.6 prevents pending Cloud Fetch downloads from delaying the last connection from closing. It also updates lz4-java, Jackson, and libthrift. Download it from the JDBC drivers archive under All JDBC Driver Versions (Simba).
The managed Xero connector is now available in Beta in Databricks Lakeflow Connect. Workspace admins can enable it from the Previews page. Use the connector to ingest accounting data such as invoices, payments, and contacts from Xero into Databricks. The connector uses OAuth user authorization, ingests transaction tables such as invoices and payments incrementally, and removes deleted records for…