Embedding Cost Calculator
Price embedding a dataset and the vector storage it will need.
Building a RAG or search index means embedding every chunk of your corpus, and that has a bill. This embedding api price estimator turns your dataset into a number: choose an embedding model, say how many texts you have and the average tokens each, and it multiplies the total tokens by the model's rate. It also estimates raw vector storage from the dimension count, so you can see the openai embeddings cost and the disk footprint in one place before you kick off an indexing run.
Your embedding job
Embedding cost
$0.5120
25,600,000 tokens on text-embedding-3-small
- Total tokens
- 25,600,000
- Vector storage
- 585.9 MB
- Dimensions
- 1536
- Price / 1M
- $0.020/1M
Prices updated January 2026. Embeddings bill on input tokens only. Storage is a raw float32 estimate (1536 dims × 4 bytes × vectors) — a real vector database adds index overhead. Verify current rates with the provider.
How it works
- 1
Pick the embedding model
Choose the model you will call. The per-million-token rate and the model's dimension count fill in underneath, so the storage math tracks the right vector size.
- 2
Describe your corpus
Enter how many texts or chunks you will embed and the average tokens per text. A 200-word passage is roughly 260 tokens; drag the slider to size a small test or a multi-million-vector index.
- 3
Read cost and storage
The headline is the embedding API cost. The stats show total tokens and the float32 storage estimate, so you can judge both the one-time spend and the ongoing database size.
Instant & 100% private — nothing is uploaded
Every calculation runs locally in your browser. The prompts, token counts and numbers you enter stay on your own device and are never sent to a server — nothing is stored, logged or shared.
Frequently asked questions
- How is the vector embedding pricing calculated?
- Embeddings are billed on input tokens only, so the cost is total tokens (texts × average tokens each) multiplied by the model's per-million-token rate. There is no output charge because an embedding returns a fixed-size vector, not generated text.
- How much storage will the vectors take?
- Each vector is its dimension count times four bytes in float32. A 1536-dimension embedding is about 6 KB per vector, so a million of them is roughly 6 GB before index overhead. Quantizing to int8 or using a lower-dimension model cuts this sharply.
- Do I pay again to search the embeddings?
- Embedding the query at search time costs a few tokens per query, which is tiny. The one-time cost here is indexing your corpus; the ongoing cost is the query embeddings plus whatever your vector database charges to store and serve them.
- Why does re-embedding cost the same again?
- Switching embedding models means re-embedding the whole corpus, because vectors from different models are not comparable. Factor that in before you commit to a model — the migration cost is the full indexing cost over again.
Important
For planning and estimates only. Prices come from a published rate table dated on the page; providers change pricing without notice, and token counts here are approximations. Confirm against the provider’s own pricing before you budget or commit.
More tools
More from the Hivly network
Free sister tools on our other sites.
