Embedding Cost Calculator
Calculate the cost of embedding your documents for RAG or semantic search, and compare rates across providers.
Every document you embed for a RAG pipeline or semantic search index gets converted into a vector by an embedding model, and that conversion is billed per token just like a chat completion — except embeddings have no separate output cost, since the model produces a fixed-size vector rather than generated text. This calculator estimates the total cost of embedding a document set given your document count and average length, and compares that cost across the major embedding providers so you can see which is cheapest before committing to one at scale.
A typical paragraph runs roughly 100–150 tokens; a full page is often 500–800.
Total Cost
$0.1000
Total Tokens
5,000,000
Cost / 1,000 Docs
$0.0100
| Model | Provider | Rate (1M tokens) | Total Cost |
|---|---|---|---|
| DeepSeek Embed | DeepSeek | $0.01 | $0.0500 |
| OpenAI Embedding v4 Small | OpenAI | $0.02 | $0.1000 |
| Cohere Embed v4 | Cohere | $0.10 | $0.5000 |
| OpenAI Embedding v4 Large | OpenAI | $0.13 | $0.6500 |
| Gemini Embedding | $0.15 | $0.7500 |
How to Use This Calculator
1. Enter your document count
The total number of documents, chunks, or passages you plan to embed.
2. Enter average tokens per document
Use a real average from your dataset if you have one; typical passages run 100–800 tokens.
3. Choose an embedding model
Select a provider to see its total cost, or scan the comparison table below for the cheapest option.
4. Read your total cost and per-1,000-document rate
Use the per-1,000-document figure to estimate cost at a different scale without re-entering numbers.
How It Works
(document_count × avg_tokens_per_doc / 1M) × rate_per_million10,000 documents at 500 tokens each is 5M tokens. At $0.02/1M tokens, that's a $0.10 total embedding cost — a one-time cost per document, not a recurring one, unless the underlying documents change.
Frequently Asked Questions
Why don't embedding models charge separately for input and output?
An embedding model doesn't generate text — it converts input text into a fixed-size numerical vector. There's no output token stream to bill separately, so embedding pricing is a single per-token rate applied to the input only.
Is embedding a one-time cost or a recurring one?
Embedding a document costs once, at the time you embed it. It becomes a recurring cost only if you re-embed the same content repeatedly (for example, re-indexing your entire dataset on every deploy instead of only embedding new or changed documents).
Do I need to re-embed my documents if I switch embedding models?
Yes. Vectors from different embedding models are not compatible with each other — switching models means re-embedding your entire document set and rebuilding your vector index from scratch.
How does chunk size affect embedding cost?
Smaller chunks mean more chunks for the same total text, and each chunk carries a small fixed overhead in addition to its content tokens. Very small chunking strategies can noticeably increase total token count compared to fewer, larger chunks covering the same source text.
Common Mistakes
❌ Re-embedding unchanged documents on every run
Embedding is billed per token every time it runs. Cache embeddings for documents that haven't changed instead of regenerating them on every indexing job — most vector databases are built around exactly this assumption.
❌ Ignoring chunking overhead
Splitting one long document into many small chunks increases total token count slightly (due to per-chunk overhead) and significantly increases the number of vectors stored, which affects vector database cost more than embedding cost — see the Vector Database Cost Calculator.
❌ Assuming the most expensive embedding model is always most accurate for retrieval
Embedding quality for retrieval doesn't scale with price as directly as generation quality does. Benchmark retrieval accuracy on your own data before assuming the priciest option is worth the cost.
Related Tools
AI Token & API Cost Calculator
Calculate precise token usage and API costs for GPT-4o, Claude Opus, Gemini, and DeepSeek. Compare billing across models in real time with a live cost breakdown.
Vector Database Cost Calculator
Estimate the monthly cost of storing and querying your vectors across Pinecone, Weaviate, and Qdrant. Compare storage and query costs by vector count and dimension.