What costs belong in a RAG pipeline?
Include document parsing, embeddings, vector storage and queries, reranking, LLM input and output, observability, data transfer, and maintenance.
Free · Private · Editable assumptions
RAG Pipeline Cost Calculator
Estimate RAG ingestion, embedding, vector database, retrieval, LLM generation, and monthly operating costs.
01 · Pipeline inputs
Separate build cost from run cost
02 · Decision view
One-time embedding cost
$0.160Monthly operating cost
$214.5Cost per 1,000 answers
$2.15LLM
$79.5Vector
$35Ops
$100
$79.5Vector
$35Ops
$100
Methodology
Inspectable by design
ingestion = document tokens × embedding rate
monthly = LLM query cost + vector store + monitoring & operationsThis planning model excludes parsing/OCR, reranking, data transfer, cache effects, provider minimums, evaluation runs, and labor outside the entered operations allowance. Use invoice or benchmark data when available.
Questions people ask
Calculator FAQ
Is embedding a monthly cost?
Initial embedding is usually an ingestion cost. Re-embedding changed documents creates recurring spend, which you can model by entering the monthly changed-document volume.
What usually drives RAG cost?
At scale, answer-generation tokens often dominate. At low volume, vector infrastructure minimums and engineering operations can matter more.
Does lower cost mean a better RAG design?
No. Retrieval quality, answer accuracy, latency, security, and evaluation matter. Cheap retrieval that produces unsupported answers is not economical.