Standard and Dedicated access modes

Standard and Dedicated access modes is the current Databricks name, previously Shared / Single user access modes (since March 2025).

Serverless pools → No isolation shared access mode → High Concurrency clusters → Shared / Single user access modes → Standard and Dedicated access modes

Compute access modes that determine Unity Catalog support, language availability, and whether a cluster is shared.

• 'Shared' became 'Standard' and 'Single user' became 'Dedicated' in 2025 - then 'Dedicated' clusters were promptly allowed to be shared with a group, making the new name ironic on arrival. 🔗
• The access mode picker now defaults to 'Auto', which decides for you from the runtime - Standard, unless you pick an ML runtime, a GPU instance type or a runtime below 14.3, in which case you get Dedicated. 🔗

⚠ Limitations: On Standard access mode, Databricks Runtime for ML, GPU compute, R, RDD APIs, spark-submit tasks, Hive UDFs, the Scala SparkContext, and streaming socket sources aren't supported; compute runs as a low-privilege user with restricted filesystem access and can't reach the instance metadata service or other services in the Databricks VPC. 🔗

Category

Compute / BI

In use from

March 2025🔗

Announced at

What's new with Unity Catalog compute, March 2025🔗

Also known as

Shared cluster, Single user cluster, shared access mode, single-user access mode

Verified

2026-08-20