What's new in ML Runtimes version 2026.08.1

Review the major features and updates for ML Runtimes.

New Features/Improvements

  • Data Connections support on Hardened runtimes: Data Connections are now fully supported on Python 3.11 and Python 3.14 Hardened runtime variants, allowing secure data connectivity for hardened runtime environments.
  • Ranger RAZ client integration (raz-client): The raz-client package has been added to Python 3.14 Hardened runtime variant.
  • R line version upgrade: Upgraded the R runtime line to support newer language features and library compatibility. Added PBJ Workbench R 4.6 Standard as the default standard R runtime.

Decommissioned and Deprecated Runtimes

The following runtime variants have been decommissioned in this release:

  • Freshline runtime variants decommissioned:
    • PBJ JupyterLab Python 3.11 Freshline
    • PBJ JupyterLab R 4.5 Freshline
  • Standard R runtime decommissioned:
    • PBJ Workbench R 4.5 Standard (replaced by PBJ Workbench R 4.6 Standard)

Fixed Issues

  • Data connection failure on Hardened runtimes (DSE-54787)

    Resolved an issue where Kerberos-authenticated data connections failed with a Cryptosystem internal error on Hardened runtimes. Kerberos authentication now properly negotiates FIPS-compatible encryption types (aes256-cts-hmac-sha384-192 and aes128-cts-hmac-sha256-128) in FIPS environments. Standard runtimes continue operating normally.