Data and AI releases

This month1,990Companies171

1,990 this month · 171 companies

External writes to managed Delta tables are now generally available

GA11 days agoUnverifiedAdded Oct 17, 2026

Support for creating and writing to Unity Catalog managed Delta tables from external Apache Spark clients is now generally available. Tables must have catalog commits enabled. See Access Databricks tables from Delta clients .

Topics: Governance, Observability

Databricks's release notes

More Unity Catalog releases

Every Unity Catalog release
GAunverified

Shared output in clean rooms is now generally available

Clean room runs can now write to a shared output schema that every collaborator can read as soon as each table is created, even before the run finishes. Use it to share results, logs, and debugging information with all collaborators. See Share output with all collaborators .

GAunverified

Create clean rooms with identity resolution providers in one step (GA)

You can now create a ready-to-use packaged clean room with an approved identity resolution provider, such as Acxiom, Adstra, LiveRamp, or The Trade Desk, in one step from the Clean Rooms page. The provider is preconfigured as the package provider, so you can add your data and run the provider's analysis right away.

GAunverified

Transactions support both managed Delta Lake and Apache Iceberg tables

Transactions are generally available for both Unity Catalog managed Delta Lake tables and Unity Catalog managed Apache Iceberg tables. Group multiple SQL statements across one or more managed Delta Lake or managed Apache Iceberg tables into a single atomic transaction using BEGIN ATOMIC ... END; or BEGIN TRANSACTION; ... COMMIT; syntax.

Betaunverified

Governed tag propagation is in Beta

Tag propagation in Unity Catalog is in Beta . When you enable tag propagation for a governed tag, Unity Catalog uses lineage to automatically apply that tag to derived tables and views whenever a supported operation runs, such as CREATE TABLE AS SELECT , CREATE VIEW , INSERT INTO ... SELECT , or MERGE .

Also shipped on Oct 13, 2026

Databricks in October 2026
Betaunverified

What's included:

- Three training methods: Supervised fine-tuning (SFT) , direct preference optimization (DPO) , and reinforcement learning with group relative policy optimization (GRPO) . Each method trains a low-rank adapter (LoRA) or the full model weights. Custom loss functions : Train with any objective you compute in PyTorch from the model's log-probabilities.