RAG Cost Estimator
One-time embeddings, monthly retrieval + generation, and a yearly roll-up
Prices are approximate list rates — verify with your provider.
Embed and generator rates are public USD-per-1M-token list prices and change. Caching, batch APIs, rerankers, vector-DB hosting, and evals are not included.
1 Corpus & chunking
Files or pages you will index
~750 words ≈ 1,000 tokens
Typical RAG chunks are 256–1,024 tokens
Overlap increases embed tokens; 10% is a common default
2 Embed model & cadence
Filled by the dropdown; edit for a custom or discounted rate
How often the full corpus is re-embedded after the initial ingest
3 Queries & generator
Question text sent to the embed model
Instructions + user question in the generator window