Google Gemini 2.5 Pro Model Retired in Databricks

Use cases

Model retirements require updating applications to prevent production failures. Use this SQL query to identify which endpoints and users still depend on a discontinued model.

How to use

Run the following query in Databricks SQL to locate usage of the old model:

SELECT
   eu.requester,
   se.endpoint_name,
   se.entity_name,
   COUNT(*) AS request_count,
   SUM(eu.input_token_count) AS total_input_tokens,
   SUM(eu.output_token_count) AS total_output_tokens,
   MIN(eu.request_time) AS first_request,
   MAX(eu.request_time) AS last_request
 FROM system.serving.endpoint_usage eu
 JOIN system.serving.served_entities se
   ON eu.served_entity_id = se.served_entity_id
 WHERE LOWER(se.entity_name) LIKE '%gemini-2.5-pro%'
 GROUP BY eu.requester, se.endpoint_name, se.entity_name
 ORDER BY request_count DESC

What happens in this code:

1. Joins the system schema tables for endpoint usage and served entities.
2. Filters by the discontinued model name across endpoints.
3. Groups by the requesting user and endpoint name to consolidate calls.
4. Returns the request count, total tokens, and usage time range.

Limitations and requirements

The Gemini 2.5 Pro model is no longer accessible for new workloads or workspaces without prior usage. The recommended replacement options are Gemini 3.1 Pro and Gemini 3.5 Flash.

Using it in a pipeline

Identify affected endpoints using the audit system, update the endpoint configuration to point to the new model, and validate application calls.