There is quite a lot of retirement / deprecation alerts related to Power BI. One of them was a bit quiet, however, Tenant Admins should be notified about these changes to prepare their organization for what’s about to come. I announced it in our company, together with the solution, and now I am sharing it also with you.
What is it about?
It wasn’t announced in Power BI Blog, but if you go to Deployment Pipelines documentation, you will find this message. It says:
“Beginning February 12, 2026, Microsoft Fabric deployment pipelines will retire support for semantic models that have not been upgraded to Enhanced Metadata.”
Now, depending how long ago your Power BI journey started, you might ask what is this Enhanced Metadata. This is a metadata format, introduced around mid 2020, built around TOM (Tabular Object Model). It helped to properly expose details related to data sources and Power Query M transformations, that were very difficult to extract before, making CI/CD scenarios very painful. New format unlocked full potential of XMLA read/write operations, allowing external tools like Tabular Editor to fully manage Semantic Models. Here is the link to feature documentation.
Enhanced Metadata is running on V3 of default Power BI data source version, and compatibility level above 1520. This is the only support format for Git integration and now will be also for Deployment Pipelines. Microsoft’s recommendation to upgrade your semantic model is simply: open your semantic model in newest Power BI Desktop and re-publish. If it’s that easy, maybe you’ve already done it? This leads to questions:
Should you be worried? Are you affected?
Short answers: yes, and not necessarily. Maybe “worried” is an exaggeration, but you should always check announcements, especially related to alerts that something can no longer work as expected. So, you probably should look into it and make sure that none of your items will be affected.
What about being affected part. Most probably you are not affected, why? Retire support is related to Deployment Pipelines. Not everyone is using it, and these people are not affected for sure. But, most importantly we should look at the Microsoft’s solution to the problem. If re-publishing helps to solve the issue, there is chance that during last 5 years, you did that at least once for all your models. If not, and you have Power BI reports older than 5 years, that didn’t require you to re-publish them with some updates – those are happy days.
However, we know already that even though you are probably safe, you should still check to be 100% sure. So, how to check it?
Three ways to check if you are affected
First option was suggested by Microsoft, other two are my recommendations. You can use them depending on your environment, as not all of them may be available to you. What Microsoft recommends is to go to Power BI Service -> right click on Semantic Model -> Check if option to Open Semanitc Model is active:
It looks fairly simple, but there are two problems with this approach:
- You must be allowed to Edit Semantic Models in Power BI Service. This may not be enabled for your Tenant or Workspace.
- Even if you feature is enabled, you still must check all models one by one -> it may take a lot of time
Second option is to check the Semantic Model Properties in Tabular Editor. Grab the xmla endpoint of your workspace -> go to Tabular Editor -> Select Model -> expand Database section -> find Compatibility Level -> scroll down and find Default Power BI Data Source Version:
Remember, we are looking to have Compatibility Level > 1520, and default data source version = PowerBI_V3. With this approach we removed one important dependency – we don’t need the option to Edit Semantic Model Online anymore. Still, the only option to check all your models is one by one. Which leads to last option.
Third approach is a natural extension of second one. We know what to check, so let’s try to automate it. We will use Notebooks in MS Fabric, together with Fabric REST Client. If you followed last article, this should be already known. What’s new this time is that we will add another package to check the Semantic Model Properties – Semantic Link Labs. This package is based on Semantic Link but provides additional functionalities on top. It is initiated by MS employee – Michael Kovalsky and has already almost 30 contributors. I created a quick script that will check all your Semantic Models in all Workspaces that you provide. To get the full script go to my Github. Main part of the code is where we get the Semantic Model Properties:
# Get semantic model definition (TMSL) and extract properties
try:
bim = get_semantic_model_definition(
dataset_id,
format="TMSL",
workspace=ws_id,
return_dataframe=False,
)
compat = bim.get("compatibilityLevel")
ds_ver = bim.get("model", {}).get("defaultPowerBIDataSourceVersion")
except Exception as ex:
compat = None
ds_ver = None
print(f"[WARN] Could not read model definition for '{dataset_name}' ({dataset_id}): {ex}")
As you can see, it’s pretty straight-forward. Especially that we must focus on two properties only. However, I really recommend you to have a close look at the entire bim return value in my code. Check the full structure of the response and see how you could retrieve other properties that might be useful to you. End result of the script is just a display of the results. It’s up to you to decide how you want to use this information further. Maybe log it somewhere, save to csv file and share with a colleague. Whatever suits your needs. In my version, you will get a simple summary:
Again, we want to see Compatibility Level above 1520, and Default Data Source Version PowerBI_V3. Code allows you to include multiple Workspace IDs in the list for iteration. I used only one to display the results for you, and it seems that this workspace is safe 🙂
Conclusion
As daunting as admin work can sometimes seem, it’s important to take every opportunity to learn something new. Whenever I dive into Semantic Link Labs, I genuinely enjoy it, thanks to the amazing work of Michael Kovalsky and the rest of the team. Be sure to explore the package — it already offers a wide range of functionality, and new features are released regularly.
Thank you for staying with me till the end, and see you in next article 🙂
