We are from FarmVaidya.ai, developing a production real-time AI voice agent for users in India.
We are evaluating Gemini Flash models through Vertex AI. For the models we are evaluating, we currently see Global and us-central1 availability, but we don’t see an India region endpoint.
Our use case is highly latency-sensitive because the application is a real-time voice conversation involving STT → Gemini → TTS. Additional network/inference latency directly affects the conversational experience.
We would like to understand:
-
Is an India region currently available or planned for these Gemini Flash models?
-
For users in India, should we use the Global endpoint or us-central1 for the lowest practical latency?
-
Is there a recommended Google Cloud architecture or routing configuration to minimize latency from India?
-
Are there upcoming India-region Gemini endpoints suitable for real-time voice applications?
-
Is there an early-access program for upcoming India-region Gemini models/endpoints?
Models we are evaluating include:
- Gemini 2.5 Flash-Lite
- Gemini 3.1 Flash-Lite
- Gemini 3.5 Flash-Lite
- Gemini 3.5 Flash
- Gemini 3.6 Flash
- Gemini 3.7 Flash
- Gemini 3.8 Flash
Company: FarmVaidya.ai
Use case: Production real-time AI voice agent
Primary users: India
Service: Vertex AI / Gemini
We would appreciate guidance from Google Cloud engineers or the community on the recommended architecture for minimizing latency for India-based users.