I want to Enable Databricks ML runtimes in a workspace.

ROMEO PEAY 20 Reputation points
2026-06-07T15:09:36.5233333+00:00

We need to enable Databricks ML runtimes (e.g., 13.3 LTS ML, 14.x ML) and GPU ML runtimes in our workspace. Currently, these options do not appear under Compute → Create Compute, and there is no Runtime Availability toggle in Settings → Compute. Please enable ML runtime feature flags for this workspace. How can I get this enabled?

Azure Machine Learning
Answer accepted by question author
Anshika Varshney 15,625 Reputation points Microsoft External Staff Moderator
2026-06-08T11:23:53.6066667+00:00

Hi @ROMEO PEAY

It sounds like you’re trying to use Databricks ML runtimes (for example, Databricks Runtime for ML versions like 13.3 LTS ML or 14.x ML) and possibly GPU-enabled ML runtimes in your workspace, but the expected options are not appearing under Compute → Create Compute, and there is no Runtime Availability option under Settings → Compute.

Based on the available documentation, here are a few things you can check:

  1. Verify required permissions

Runtime-related features may not appear if the user does not have sufficient permissions.

Please ensure that your account has the required permissions, such as the AzureML Data Scientist role (or equivalent access) on the Azure Machine Learning workspace.

  1. Validate workspace prerequisites

For Azure Machine Learning runtime creation and updates, verify the following:

  • The default datastore workspaceblobstore is configured as a Blob datastore.
  • The workspace contains workspaceworkingdirectory.

Missing or misconfigured workspace resources can affect runtime availability.

  1. Ensure runtimes are up to date

If the runtime options are partially visible but not functioning correctly, try updating the runtime from the Azure portal using the Check version or update options available on the runtime details page.

  1. Databricks Runtime for Machine Learning

If your requirement is specifically related to Databricks Runtime for Machine Learning:

  • While creating compute/cluster resources, select the Machine learning option in the Databricks compute creation flow.
  • For GPU-enabled ML runtimes, select a GPU-enabled worker/VM type.

Once ML runtimes are available in the UI, GPU support is generally configured through the selected worker type.

About enabling “feature flags”

Based on the currently available documentation, there is no publicly documented customer-facing method to manually enable specific Databricks ML runtime versions through feature flags for a workspace.

References

I hope this helps. Please share additional details or screenshots of the compute creation page if the issue still persists.

Thank you!

Was this answer helpful?

1 person found this answer helpful.
0 comments

2 additional answers

Sort by: Most helpful
Most helpful Newest Oldest
  1. ROMEO PEAY 20 Reputation points
    2026-06-09T22:13:34.7433333+00:00

    I was able to resolve the issue by configuring my workspace for Databricks runtime 13.3 LTS ML with instructions similar to the above solution.

    Was this answer helpful?

    0 comments

  2. AI answer

    2026-06-07T15:09:53.54+00:00

    Databricks ML runtimes and GPU ML runtimes are enabled and selected entirely within Azure Databricks; there is no separate “ML runtime feature flag” or Azure ML-side control to turn them on.

    To use Databricks ML runtimes in a workspace:

    1. Create a classic compute resource with Databricks Runtime ML
      • In Azure Databricks, go to Compute → Create compute.
      • Select the Machine learning checkbox. This creates a compute resource using Databricks Runtime for Machine Learning with pre-built ML/DL libraries and infrastructure.
      • This checkbox automatically sets the access mode to Dedicated and assigns a dedicated user; this can be changed in the Advanced section if needed.
      • Databricks Runtime ML versions (for example, 13.3 LTS ML) are managed as part of the runtime release train; see the Databricks Runtime ML release notes for the exact versions available in the workspace.
    2. Enable GPU ML runtimes
      • In Compute → Create compute, to create a GPU ML compute:
        • Ensure the Machine learning checkbox is checked. The GPU ML version is chosen automatically based on the worker type.
        • Uncheck Photon acceleration; Photon is not supported with GPU instance types.
        • In Worker type, select a supported GPU instance type (for example, from the NCads_H100_v5 or NC_A100_v4 families listed under supported instance types).
        • Optionally check Single node to get a single GPU instance.
      • When using the Clusters/Compute API instead of the UI:
        • If kind = CLASSIC_PREVIEW, set "use_ml_runtime": true.
        • If kind is not set, choose a GPU-enabled ML runtime by setting spark_version to a GPU ML version such as 15.4.x-gpu-ml-scala2.12.
    3. Use Databricks Runtime ML for ML workloads
      • Databricks Runtime ML provides pre-configured environments with major ML libraries and built-in GPU support (including drivers and supporting libraries) for both CPU and GPU clusters.
      • Additional libraries can be installed via compute-scoped libraries or notebook-scoped Python libraries if custom environments are required.

    If the Machine learning checkbox or ML runtimes still do not appear in the workspace’s Compute creation UI, this is a workspace/service configuration issue that must be handled by Azure Databricks support or the Azure administrator for that workspace; there is no documented self-service feature flag to enable ML runtimes beyond using the options described above.

    References:

    AI-generated content may be incorrect. Read our transparency notes for more information.

    Was this answer helpful?

Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.