Hi community,
Starting around April 21, 2026, Genie Code in our Azure Databricks workspace has severely regressed and no longer functions the way it did before. No changes were made on our end — no DBR updates, no workspace config changes.
We are a team of Data Engineers working primarily with PySpark and SQL on Azure Databricks with Unity Catalog.
What it used to do:
- Generated accurate PySpark and SQL code
- Understood notebook context and table schemas
- Gave concise, relevant suggestions
- Autocompleted code correctly
- Performed inline editing and creation of code directly in notebooks
- Completed thought processes quickly
- Provided great summaries and analysis of data and code
What it does now:
- Responses are much lower quality overall
- Gives wrong or generic code that doesn't fit the context
- No longer provides inline guidance or inline code changes
- Does not find tables, columns, or context it previously could
- Won't create or apply inline code edits in notebooks
- Makes repetitive suggestions to "run code" without actually helping
- Significantly less helpful compared to 2 days ago
Workspace: Azure Databricks (region unknown), workspace URL: adb-4894429498919227.7.azuredatabricks.net
Has anyone else experienced this recently? Is there a known model or product update that may have caused this? Any guidance from the Databricks team would be greatly appreciated.
