Skip to content

Embeddings Cost Calculator

Calculate embedding costs for your document collection. Compare pricing across OpenAI, Cohere, Voyage, and Google embedding models with chunking estimates.

FreeNo SignupNo Server UploadsZero Tracking

Select models to compare

Chunks / Doc

6

Total Chunks

6,000

Total Tokens

3,000,000

ModelProvider$/1M TokensDimensionsCost/DocTotal Cost
Doubao Embedding
VolcengineFree2,560$0.00$0.00
Doubao Embedding Large
VolcengineFree2,048$0.00$0.00
Doubao Embedding Large Text 240915
VolcengineFree4,096$0.00$0.00
Doubao Embedding Large Text 250515
VolcengineFree2,048$0.00$0.00

Cost Comparison

Doubao Embedding
$0.00
Doubao Embedding Large
$0.00
Doubao Embedding Large Text 240915
$0.00
Doubao Embedding Large Text 250515
$0.00

Token estimation assumes ~1.33 tokens per word or ~4 characters per token. Chunks may exceed model max input length for some providers; real implementations truncate accordingly. Google Gemini Embedding is free within usage limits. Prices reflect current published rates.

Export

How to Use Embeddings Cost Calculator

  1. 1

    Enter document details

    Input the number of documents, average length (in words or characters), chunk size, and overlap.

  2. 2

    Select models to compare

    Toggle embedding models on or off to compare 3 or more models side by side.

  3. 3

    Review chunking stats

    See how many chunks your documents produce and total tokens to embed.

  4. 4

    Compare costs

    View total embedding cost and cost per document for each model, sorted cheapest first.

Frequently Asked Questions

Chunks per document = ceil(document_tokens / (chunk_size - chunk_overlap)). For example, a 2000-word document (~2660 tokens) with 500-token chunks and 50-token overlap produces ceil(2660/450) = 6 chunks.

Chunk overlap means each chunk shares some tokens with the previous chunk, improving retrieval quality at chunk boundaries. Common values are 10-20% of chunk size. More overlap means more chunks and higher cost.

We estimate 1.33 tokens per word for English text (the average across major tokenizers). For character input, we use 4 characters per token. Actual counts vary by language and content.

Google offers Gemini Embedding at no cost within certain usage limits. For high-volume production use, check Google's current terms as free tier limits may apply.

For most use cases, OpenAI text-embedding-3-small offers the best balance of cost and quality. For higher retrieval accuracy, text-embedding-3-large or Voyage 3 are good choices. Cohere embed-v4 excels at multilingual content.