Serverless python version

[GeorgeFmz]

In the documentation about serverless compute release notes, it is stated that "Serverless compute for notebooks and jobs uses environment versions", and that "Serverless compute always runs using the most recently released version listed here.". At time of writing, the most recently released version is 16.3, which is described as "roughly corresponds to Databricks Runtime 16.3.". That DBR runs python 3.12, but the equivalent serverless environment is python 3.11.

What is the actual equivalence to the normal DBR releases? Are environment versions different from the versions listed there? A large percentage of our jobs can't take advantage of serverless compute because they need python 3.12, is there a roadmap on when that will be available?

Thank you in advance

[Walter_C]

May I know the region where your workspace is located? As you have mentioned Serverless should be running on 16.3 which should have the Python 3.12

[GeorgeFmz]

Thanks for the quick reply. We are in eu-central-1. We see python 3.11.10 when running the following on a notebook attached to Serverless environment version 2 (also on jobs defined with this environment):

from sys import version

version

[Walter_C]

It seems the configuration aligns with doc https://docs.databricks.com/aws/en/release-notes/serverless/environment-version-two which specifies that the Python version in 3.11.10

[GeorgeFmz]

Yes I saw this documentation too, but it is not clear to me what compute the release notes for 16.3 reference then? How could I use serverless compute on jobs and notebooks that runs that DBR version?