Tech & Energy
AI RAG Cost Calculator
Retrieval-Augmented Generation has 4 costs: embedding the corpus, storing vectors, embedding queries, and the LLM call. Add them up.
Monthly total
—
- Embedding one-time
- —
- Vector DB/mo
- —
- LLM/mo
- —
FAQ
Why does LLM dominate? +
Vector DB costs are usually $5-50/mo at small scale. LLM costs scale linearly with traffic — easily $500-5000/mo at moderate volume.
Reranking? +
Adds cost (Cohere ~$1/1k reranks) but cuts top-k context by 50%+, often net savings. Worth testing.