Databricks releases, every cloud
| Release | Date | ||||
|---|---|---|---|---|---|
| Genie Code for AI Runtime (Public Preview) Generates distributed training code and resolves environment issues. Requires AI Runtime. | On AWS, read it on their docs | On Azure, read it on their docs | Not on GCP | Unknown on SAP | Aug 27, 2026 |
| Web terminal on serverless GPU compute (AI Runtime) is in Public Preview Runs shell commands, monitors GPU usage, and manages files on serverless GPU compute. Requires environment version 5 or above. | On AWS, read it on their docs | On Azure, read it on their docs | Not on GCP | Unknown on SAP | Jul 31, 2026 |
| Protobuf tensor input for custom model serving endpoints (Public Preview) Replaces JSON with serialized KServe v2 ModelInferRequest. Requires endpoints deployed after July 9, 2026. | On AWS, read it on their docs | On Azure, read it on their docs | On GCP, read it on their docs | Not on SAP | Jul 9, 2026 |
| Custom Docker images for AI Runtime CLI workloads (Beta) Allows custom system library versions and complex dependencies, requires setup via Use custom Docker images. | On AWS, read it on their docs | On Azure, read it on their docs | Not on GCP | Not on SAP | Jun 29, 2026 |
| AI Runtime CLI (Beta) Submits and manages distributed training workloads on serverless GPU compute from a local machine. | On AWS, read it on their docs | On Azure, read it on their docs | Not on GCP | Not on SAP | Jun 12, 2026 |
| Upcoming behavior change: Choose entitlements when adding principals to workspaces Coming soon Grants entitlements explicitly when adding principals, replacing inheritance from users system group. | On AWS, read it on their docs | On Azure, read it on their docs | On GCP, read it on their docs | Not on SAP | Jun 8, 2026 |
| Run the change, do not just read it. Hands-on labs in your own Databricks workspace, graded when you submit. Browse labs | |||||
| Serve custom LLMs with Custom Model Serving (Beta) Supports multimodal models and PEFT recipes, unlike Foundation Model APIs. | On AWS, read it on their docs | On Azure, read it on their docs | Not on GCP | Not on SAP | May 20, 2026 |
| AI Runtime 1xH100 accelerator (Beta) Supports 1xH100 accelerator, see Hardware options for details. | On AWS, read it on their docs | On Azure, read it on their docs | Not on GCP | Not on SAP | Apr 21, 2026 |
| Upcoming breaking change: default behavior when deleting a Unity Catalog pipeline Coming soon Deletes now retain associated tables by default, set cascade=true to remove them. | On AWS, read it on their docs | On Azure, read it on their docs | On GCP, read it on their docs | Not on SAP | Apr 9, 2026 |
| AI Runtime is now in Public Preview Adds GPU support to serverless compute for deep learning workloads. | On AWS, read it on their docs | On Azure, read it on their docs | Not on GCP | Not on SAP | Mar 19, 2026 |
| Declarative Feature Engineering APIs (Beta) Define time-windowed aggregation features from data sources, materialize to Delta tables or Lakebase. Available in us-east-1 and us-west-2 regions. | On AWS, read it on their docs | Not on Azure | Not on GCP | Not on SAP | Mar 9, 2026 |
| Endpoint telemetry for custom model serving endpoints (Beta) Stores logs, traces, and metrics in Unity Catalog Delta tables using OpenTelemetry. | On AWS, read it on their docs | On Azure, read it on their docs | Not on GCP | Not on SAP | Mar 6, 2026 |
| Change to use_case field in VPC endpoint API responses Coming soon Changes from WORKSPACE_ACCESS to GENERAL_ACCESS, see API docs for details. | On AWS, read it on their docs | Not on Azure | On GCP, read it on their docs | Unknown on SAP | Feb 24, 2026 |
- Genie Code for AI Runtime (Public Preview)
Generates distributed training code and resolves environment issues. Requires AI Runtime.
- Web terminal on serverless GPU compute (AI Runtime) is in Public Preview
Runs shell commands, monitors GPU usage, and manages files on serverless GPU compute. Requires environment version 5 or above.
- Protobuf tensor input for custom model serving endpoints (Public Preview)
Replaces JSON with serialized KServe v2 ModelInferRequest. Requires endpoints deployed after July 9, 2026.
AWS, read it on their docsAzure, read it on their docsGCP, read it on their docsSAP(not on this cloud) - Custom Docker images for AI Runtime CLI workloads (Beta)
Allows custom system library versions and complex dependencies, requires setup via Use custom Docker images.
- AI Runtime CLI (Beta)
Submits and manages distributed training workloads on serverless GPU compute from a local machine.
- Upcoming behavior change: Choose entitlements when adding principals to workspaces Coming soon
Grants entitlements explicitly when adding principals, replacing inheritance from users system group.
AWS, read it on their docsAzure, read it on their docsGCP, read it on their docsSAP(not on this cloud) - Run the change, do not just read it. Hands-on labs in your own Databricks workspace, graded when you submit. Browse labs
- Serve custom LLMs with Custom Model Serving (Beta)
Supports multimodal models and PEFT recipes, unlike Foundation Model APIs.
- AI Runtime 1xH100 accelerator (Beta)
Supports 1xH100 accelerator, see Hardware options for details.
- Upcoming breaking change: default behavior when deleting a Unity Catalog pipeline Coming soon
Deletes now retain associated tables by default, set cascade=true to remove them.
AWS, read it on their docsAzure, read it on their docsGCP, read it on their docsSAP(not on this cloud) - AI Runtime is now in Public Preview
Adds GPU support to serverless compute for deep learning workloads.
- Declarative Feature Engineering APIs (Beta)
Define time-windowed aggregation features from data sources, materialize to Delta tables or Lakebase. Available in us-east-1 and us-west-2 regions.
- Endpoint telemetry for custom model serving endpoints (Beta)
Stores logs, traces, and metrics in Unity Catalog Delta tables using OpenTelemetry.
- Change to use_case field in VPC endpoint API responses Coming soon
Changes from WORKSPACE_ACCESS to GENERAL_ACCESS, see API docs for details.
Headlines, dates and product areas are Databricks' own, and every item links to the note it came from. The one-line summaries are ours.