TokenPad
RAG

Embedding Cost Calculator

Price an index build, its re-indexing, and every query that hits it.

Settings
Embedding CostExact
0Token change
Token change0no change
Input tokens0what you pasted
Output tokens0what you would send
Token cost of this result
Output tokens0
As input$0.00
× 100K requests$0.00

Everything on this page runs in your browser. Nothing you paste is transmitted, because there is no server here to transmit it to.

Result

Three costs, not one

People price the initial index and forget the other two. Re-indexing on a schedule multiplies that build cost by however many times a year you run it, and query embedding is a permanent per-request charge that scales with traffic rather than with corpus size.

At high query volume the third line overtakes the first two, and the fix is caching rather than a cheaper model: in most products a small set of questions accounts for most of the traffic.

Keep it in proportion

Embedding is almost always the cheap part of a RAG system. The generation call that consumes the retrieved chunks typically costs an order of magnitude more, because generation models are priced far above embedding models.

Use this to sanity-check the index, then spend your attention on the retrieval budget, which is where the money actually is.

Frequently asked questions

How often should I re-index?
Incrementally, when documents change, rather than on a schedule. Full rebuilds are usually a workaround for not tracking which documents changed, and they are the most expensive way to solve that problem.
Do larger embedding dimensions cost more to generate?
Not usually — the price is per token, not per dimension. Dimensions affect storage and search cost instead, which the vector storage estimator covers.

More rag tools