Requesting Quota Increase & Guidance for Scaling Gemini API to 10,000+ Users

Google AI Studio
1

Dear Google Cloud and Gemini Support Team,

I hope this email finds you well.

I have built and deployed a production-ready application powered by the Gemini API on Google Cloud Run. We currently have over 500 active users, but we are experiencing critical performance issues: when concurrent usage exceeds 20 active users, the application hangs, requests queue up, and the service becomes unresponsive.

We are targeting a rapid scale-up to support **10,000+ total users** and handle at least **1,000+ concurrent active users** smoothly.

Please note that we have a **dedicated monthly budget of up to $1,500 USD** allocated specifically for Google Cloud infrastructure and Gemini API usage. We are fully prepared to transition to premium, paid, or enterprise-tier services immediately to guarantee 99.9% availability and ultra-low latency.

To help us leverage our budget effectively, could you please guide us on the following:

  1. Premium API Tiers & Vertex AI Quota Upgrades:
    How can we upgrade our Gemini API limits (specifically Requests Per Minute [RPM] and Tokens Per Minute [TPM]) to safely handle 1,000+ concurrent users? Should we transition to Vertex AI with dedicated/provisioned throughput, and how can we request the necessary quota increases?

  2. Enterprise-Grade Cloud Run Configuration:
    With our budget, what are the recommended configurations for Cloud Run? Specifically, should we enable “Always-on CPU” (disabling CPU throttling), set a higher “Minimum Instances” count (e.g., min-instances = 3 or 5) to eliminate cold starts, and what is the optimal CPU/Memory specification per container?

  3. High-Concurrency Architectural Patterns:
    Does Google recommend implementing specific middleware (like Memorystore/Redis for caching, or Cloud Tasks for request queuing) to prevent the application from hanging when calling Gemini APIs under extreme concurrent loads?

We are ready to implement these upgrades immediately. Please let us know the best path forward, or if we can be connected with a Google Cloud Solutions Architect to expedite this transition.

Thank you for your prompt assistance.