DevKits

Embedding Cost Calculator (OpenAI, Voyage, Cohere, Gemini)

Estimate the cost of embedding a corpus — pick a model, enter the number of documents and average tokens each, and get the total token count and dollar cost. 100% local.

Last updated:

Comments

Pick an embedding model, enter how many documents you're embedding and the average tokens each, and get the total token count and dollar cost — before you kick off the job. Estimates run in your browser.

Total tokens

25,600,000

Vectors produced

100,000

Estimated cost

$0.5120

0 chars · 1 lines

Embedding APIs bill only input tokens. Token estimates run in your browser (±3%). Prices are per 1M tokens as of the model list; verify with the provider before large jobs.

What is Embedding Cost Calculator?

Embedding a corpus for RAG or semantic search is usually a one-time (or periodic) batch that can quietly cost a lot when you have millions of chunks. Unlike chat models, embedding APIs bill only the input tokens — there's no output charge — so the cost is simply total tokens × the model's per-token rate. This calculator lets you compare models (text-embedding-3-small vs 3-large, Voyage, Cohere, Gemini), plug in your document count and average tokens per chunk, and see the total spend, so you can budget the job and pick the cheapest model that meets your quality bar.

How to estimate embedding cost

  1. 1Choose an embedding model.
  2. 2Enter the number of documents/chunks and average tokens each (or paste a sample chunk to estimate).
  3. 3Read the total token count and dollar cost.

Key Concepts

Embedding
A vector representation of text used for semantic search and RAG.
Chunking
Splitting documents into embeddable pieces — affects both count and quality.

Frequently Asked Questions

Do embeddings charge for output tokens?

No. Embedding APIs bill only the input tokens you send; the output vector is free. That's why this calculator only asks for input token counts.

How do I know my average tokens per document?

Paste one representative chunk and click 'Use sample's token count' to fill the field with an estimate, or enter a known value directly.

Are the prices exact?

They reflect each provider's published per-1M-token rate at the time of the model list. Verify with the provider before a large one-off job.

Try Next

Cosine Similarity

Compute cosine similarity, dot product, and Euclidean distance between two vectors online. Perfect for debugging LLM embeddings, semantic search, and RAG pipelines. 100% local — vectors never leave your browser.

Related Tools

Reference & Guides