Scaling RAG from MVP to 15M Legal Docs – Cost & Stack Advice

Gemini API
, ,
1

Hi all;

We are seeking investment for a LegalTech RAG project and need a realistic budget estimation for scaling.

The Context:

  • Target Scale: ~15 million text files (avg. 120k chars/file). Total ~1.8 TB raw text.

  • Requirement: High precision. Must support continuous data updates.

  • MVP Status: We achieved successful results on a small scale using gemini-embedding-001 + ChromaDB.

Questions:

  1. Moving from MVP to 15 million docs: What is a realistic OpEx range (Embedding + Storage + Inference) to present to investors?

  2. Is our MVP stack scalable/cost-efficient at this magnitude?

Thanks!

3

Hi @Ftrea ,

Thanks for your question. While I’d love to help, this forum is dedicated to the Gemini API and Google AI Studio. For issues with other Google products, it’s best to check their specific help communities or support pages.

5

I wanted to ask about the estimated cost for my RAG system using gemini-embedding-001.