Databricks on GCP
1090 releases that have landed on GCP, newest first.
- Discover page and domains (Beta)
Groups data assets by functional area, built on governed tags.
- Genie Agent (formerly Genie Space) Agent mode rename and improvements
Renamed from Research Agent, now produces faster and more concise responses for what-type questions.
- Scoped personal access tokens (Beta)
Limits permissions by selecting token type and API scopes, replacing legacy tokens.
- TikTok Ads connector (Beta)
Ingests data from TikTok Ads into Databricks.
- Anthropic Claude Sonnet 4.6 now available as a Databricks-hosted model
Accessible via Foundation Model APIs on a pay-per-token basis.
- Databricks Runtime 18.1 and Databricks Runtime 18.1 ML are in Beta
Powered by Apache Spark 4.1.0.
- Run the change, do not just read it. Hands-on labs in your own Databricks workspace, graded when you submit. Browse labs
- Qwen3-Embedding-0.6B now available in Public Preview as a Databricks-hosted model
Accessible via Foundation Model APIs on a pay-per-token basis.
- AI Gateway (Beta)
Replaces Unity Gateway, adds rich UI and improved observability.
- Dashboard enhancements and improvements
Supports EXCLUDE syntax on multiple fields, HTML cell styles, and cross filtering in pivot tables.
- File events enabled by default on new external locations
Enabled on new locations, detects file changes for storage jobs and pipelines, can be disabled after creation.
- Genie Agent (formerly Genie Space) enhancements and improvements
Genie now generates heatmap and sankey visualizations and increases table limit to 30.
- HubSpot connector (Beta)
Ingests HubSpot Marketing Hub data into Databricks via Lakeflow Connect.
- Configure default Python package repositories for Spark Declarative Pipelines on Lakeflow (Public Preview)
Allows installation from internal Python repositories without index-url or extra-index-url. Requires workspace admin configuration.
- Databricks SQL version 2025.40 is now available in Preview
Adds SQL scripting with procedural logic and expands parameter markers and IDENTIFIER clause support.
- New documentation for authoring and deploying agents on Databricks Apps
Supports LangGraph, PyFunc, and OpenAI Agent SDK.
- Serverless egress control is now generally available
Manages outbound connections with restricted access and FQDN filtering.
- Deploy Databricks apps from Git repositories (Beta)
Deploys from any branch, tag, or commit, without uploading files to the workspace.
- Query tags for SQL warehouses (Public Preview)
Apply custom tags to SQL workloads for cost attribution and filtering. Tags can be set via session parameters or SQL statement.
- Get this table by email. One Monday mail covering the week, filtered to the clouds and products you run. Weekly digest
- Anthropic Claude Opus 4.6 now available as a Databricks-hosted model
Accessible via Foundation Model APIs, supports pay-per-token and query reasoning and vision models.
- Applying filters, masks, tags, and comments to pipeline-created datasets is now GA
Modifies ETL and ingestion pipelines using CREATE, ALTER, or Lakeflow UI. Applies to Spark Declarative Pipelines on Lakeflow and Lakeflow Connect.
- Dashboard enhancements and new capabilities
Parameterized dashboard tasks use job parameters to set filter values. Genie is on by default for published dashboards.
- Default SQL warehouse settings (General Availability)
Sets default warehouse for SQL editor and other surfaces, can be overridden by users.
- Genie Agent (formerly Genie Space) improvements
Embedded compute credentials are now included for all users. Column lists are paginated.
- Genie One app thumbnail support
Uploads replace default icons, require PNG format.
- Google Ads connector (Beta)
Ingests data from Google Ads into Databricks via Lakeflow Connect.
- Connect Databricks Assistant to MCP servers
Connects to external tools and data sources through MCP servers, requiring permission to use added servers.
- Select tables and create pivot tables in Google Sheets
Imports data via Databricks Connector, selecting from Catalog Explorer.
- Automatically provision users (JIT) GA
Creates accounts during SSO login if none exist. Requires single sign-on setup.
- Data Quality Monitoring Anomaly Detection public preview
Enabled at schema level, learns from historical data patterns, and consolidates health into a system table.
- New GCP regions support serverless compute
Available in me-central1 and southamerica-east1 regions.
- Regional model hosting for Genie Agents in Japan
Genie Agents in asia-northeast1 use locally hosted models, eliminating cross-Geo processing.
- Tag Databricks Apps (Public Preview)
Organize and categorize apps, but search is not supported.
- Zendesk Support connector (Beta)
Ingests ticket, help center, and community forum data from Zendesk Support.
- Lakeflow jobs system tables are now generally available (GA)
Provide insights into all jobs and job runs in your account.
- Paste images into notebooks
Copies images from local file system using ⌘ + V or Ctrl + V.
- Trigger on update for pipelines (GA)
Refreshes pipeline when source table changes, usable in Databricks SQL pipeline schedules.
- View and restore your recent workspace tab sessions
Restores up to 10 recent sessions.
- Dashboard visualization and data improvements
Widgets render pivot tables at full width, and download limits increase to 10,000x1000 rows and columns.
- Genie Agent (formerly Genie Space) API and user experience improvements
Supports downloads up to 1 GB, fixes chart generation, and adds collapsible error messages.
- OpenAI GPT-5.2 Codex now available as a Databricks-hosted model
Excels at code generation and refactoring tasks. Accessible via Foundation Model APIs pay-per-token.