Pricing
Vector RAG is billed from credits — embeddings for documents you add and searches you run.
Vector RAG has no separate subscription. Usage is billed from your credits — the same balance used across Inference.
What you pay for
| Activity | Billed as |
|---|---|
| Adding documents | Embedding tokens for every chunk that is embedded |
| Running a search | Embedding tokens to embed the question |
| Syncing a connected source | Embedding tokens for every chunk fetched and embedded |
Storage of embedded documents is included; you are charged for the embedding work itself, priced per token at the same rates as inference. See Inference pricing for the per-model token rates.
Note
Top up and track your balance under credits. Each collection shows the tokens it has consumed so you can attribute cost.