The following tables summarize the feature support across MATLAB releases and underlying platform versions. If exceptions to the following constraints are encountered, please contact MathWorks databricks@mathworks.com.
| Workflows | R2026b | R2026a | R2025b | R2025a | R2024b | Toolbox Requirements |
|---|---|---|---|---|---|---|
| REST APIs | Yes | Yes | Yes | Yes | Yes | |
| JDBC | Yes | Yes | Yes | Yes | Yes | Database Toolbox |
| ODBC | Yes | Yes | Yes | Yes | Yes | Database Toolbox |
| Databricks Connect (v2) | Yes | Yes | Yes | Yes | Yes | |
| Libraries | Yes | Yes | Yes | Yes | Yes | Compiler & Compiler SDK |
| UDFs | Yes | Yes | Yes | Yes | Yes | Compiler & Compiler SDK |
| Unity Catalog | Yes | Yes | Yes | Yes | Yes | |
| Notebook Tasks | Yes | Yes | Yes | Yes | Yes | |
| MATLAB Tasks | Yes | Yes | Yes | Yes | Yes | |
| Compiled Simulink [N1] | Yes | Yes | Yes | Yes | Yes | Simulink, Compiler, Compiler SDK & Simulink Compiler |
| Spark Jar Tasks | Yes | Yes | Yes | Yes | Yes | Compiler & Compiler SDK |
| Spark Support | ||||||
| Spark 3 - Compiled | Yes | Yes | Yes | Yes | Yes | Compiler & Compiler SDK |
| Spark 3 - DB Connect | Yes | Yes | Yes | Yes | Yes | |
| Databricks Runtimes | ||||||
| 18.x LTS | Yes | Yes | Yes | Yes | Yes | |
| 17.3.x LTS | Yes | Yes | Yes | Yes | Yes | |
| 16.4.x LTS | Yes | Yes | Yes | Yes | Yes | |
| 15.4.x LTS | Yes | Yes | Yes | Yes | Yes | |
| 14.3.x LTS | Yes | Yes | Yes | Yes | Yes | |
| 13.3.x LTS | Yes | Yes | Yes | Yes | Yes |
| Notes | |
|---|---|
| [N1] | Requires a Linux® based development system, see below. |
For a given supported version of desktop MATLAB the same release of the MATLAB
runtime must be used to run deployed code that has been compiled as a .whl file.
Runtimes are typically configured at setup time and can be freely downloaded
from: https://www.mathworks.com/products/compiler/matlab-runtime.html.
With MATLAB® Compiler SDK™ workflows where a Python .whl file is produced,
the Python version used by the Databricks runtime must be a version that is supported
by the release of MATLAB used to compile the .whl file to be deployed to Databricks.
Versions of Python compatible with MATLAB releases: https://www.mathworks.com/support/requirements/python-compatibility.html
| MATLAB Release | Supported Python versions (Compiler SDK) |
|---|---|
| R2026b | 3.9, 3.10, 3.11, 3.12, 3.13 |
| R2026a | 3.9, 3.10, 3.11, 3.12, 3.13 |
| R2025b | 3.9, 3.10, 3.11, 3.12 |
| R2025a | 3.9, 3.10, 3.11, 3.12 |
| R2024b | 3.9, 3.10, 3.11, 3.12 |
Python versions used by supported Databricks runtimes are as follows, note patch version changes are to be expected:
| Databricks runtime | Python version | Cluster OS |
|---|---|---|
| 18.x LTS | 3.12 | Ubuntu® 24.04.2 LTS |
| 17.3.x LTS | 3.12 | Ubuntu 24.04.2 LTS |
| 16.4.x LTS | 3.12 | Ubuntu 24.04.2 LTS |
| 15.4.x LTS | 3.11.0rc1 | Ubuntu 22.04.4 LTS |
| 14.3.x LTS | 3.10.12 | Ubuntu 22.04.4 LTS |
| 13.3.x LTS | 3.10.12 | Ubuntu 22.04.4 LTS |
All supported MATLAB releases (R2024b and later) are compatible with the Python versions used by these Databricks runtimes.
See also: https://docs.databricks.com/en/dev-tools/sdk-python.html.
In general this package aims to support Databricks LTS releases and is developed and tested with those in mind. Non LTS releases can normally be used too, however this may require some customization, e.g. building a custom Databricks Connect library or in the cluster creation process.
Setup supports Databricks LTS runtimes up to and including 18.x LTS.
A container registry hosted in the same data center as the Databricks instance is required for:
- The MATLAB Desktop environment on Databricks.
- MATLAB runtimes on Databricks clusters using Databricks runtimes 17.x and later.
- MATLAB R2025a and later runtimes using Databricks runtime 16.x without internet access.
The required Docker® files are provided. See also: Container Services.
Python 3.10, 3.11 or 3.12 are required to build the Databricks Connect libraries.
The Python environments must have pip and venv or virtualenv installed.
-
Databricks cluster nodes always use Linux. The workflows above are cross platform on supported MATLAB platforms (Windows®, Linux & macOS), where not otherwise noted.
-
Simulink Compiler usage involves native code generation and thus the OS of the development platform must match that of the deployment platform i.e. Linux, see [N1]. Typically the best option is to use the exact same OS distribution and version as used by the Databricks runtime which is used. This should avoid
glibcand any other library version conflicts. At the very least make sure the development system is not using a newerglibcversion than the runtime image. Check the Databricks runtime releases documentation to see which exact base OS is used by the different Databricks runtime versions. -
Use on macOS will require additional tools e.g. Bash.