Hi,
Could someone clarify the retirement status of Gemini 2.5 Flash-Lite? There currently seems to be conflicting information across the documentation.
The Gemini API deprecations page lists gemini-2.5-flash-lite as having “No shutdown date announced”:
However, the Gemini Enterprise / Vertex AI model lifecycle page currently lists a retirement date of October 20, 2026:
Could you clarify whether:
-
The different dates/statuses are simply a documentation discrepancy; or
-
Gemini 2.5 models now have different retirement policies depending on whether they are accessed through the Gemini Developer API or Vertex AI / Gemini Enterprise?
For context, I currently use Vertex AI batch prediction with a fine-tuned Gemini 2.5 Flash-Lite model, so I specifically need to know whether my fine-tuned model/batch workloads are still expected to stop working in October 2026, or whether the “until further notice” guidance also applies to Vertex AI.
One additional concern for our migration planning is that the current batch inference documentation states that batch inference is not supported for tuned Gemini 3 and higher models, which means there does not appear to be a like-for-like migration path for our current tuned 2.5 Flash-Lite + batch workflow
Thanks!