We use Gemini 2.5 flash with location set to Australia Southeast via vertex AI because we need local storage and inference. 2.5 flash is set for shutdown in October 2026 but Google doesn’t currently offer an alternative model served on the regional endpoint.
Any advice?