Loading page
About this tool
Estimate AI cost per document from pages, words per page, chunk overlap, OCR and embedding rates.
The Cost Per Document Processed Calculator turns pages and words per page into tokens using the standard English estimate of about 0.75 words per token (roughly four characters), then adds everything a real pipeline bills for: chunk overlap repeated between neighbouring chunks, instruction tokens sent with each chunk, extra verification passes, OCR per page and embedding for search. It reports cost per document, per page, per thousand documents and per month, with a breakdown showing which line dominates. Useful for anyone pricing invoice extraction, contract review, claims processing or a RAG ingest.
Open Cost Per Document Processed Calculator on AltFTool — it loads instantly in your browser.
Enter the values you already know.
Fine-tune the options to match your scenario.
Read the result and use it in your planning or reporting.
Counts overlap, per-chunk prompts and multiple passes, not just the raw document tokens.
Shows what share comes from input, output, OCR and embedding, so you optimise the right line.
Per document, per page, per thousand and per month from one set of inputs.
For ordinary English prose, about 0.75 words per token means a 500-word page is roughly 667 tokens, so a 12-page document lands near 8,000 tokens. Dense tables, code and most non-English scripts use noticeably more tokens per word, so measure a real sample with your tokeniser before pricing at scale.
It depends on length and pipeline, not just the model. A 12-page document chunked at 2,000 tokens with 200-token overlap, a 300-token instruction per chunk and a 700-token structured output costs about 0.04 in model calls at 3 and 15 per million tokens — with OCR at 0.0015 a page the total is closer to 0.059.
Yes, directly. Every overlap window is billed again as part of the next chunk, so a document split into four chunks with 200-token overlap pays for 600 extra input tokens on each pass. Overlap improves retrieval quality at chunk boundaries, so the question is whether that quality is worth the percentage it adds to your input bill.
Attack the largest line first — this calculator shows which one it is. Common wins are cutting overlap, moving the verification pass to a cheaper model, shortening the per-chunk instruction, embedding once rather than on every reprocess, and skipping OCR for documents that already contain a text layer.
Add the Cost Per Document Processed Calculatorwidget to your blog or website — free, responsive, no signup. Just keep the “Widget by AltFTool” credit link visible.
<iframe src="https://www.altftool.com/embed/widget/cost-per-document-processed-calculator"
title="Cost Per Document Processed Calculator — free AltFTool widget"
width="100%" height="640" style="border:0;border-radius:12px;overflow:hidden"
loading="lazy" referrerpolicy="no-referrer-when-downgrade"></iframe>
<p style="font-size:12px;margin:4px 0 0">Widget by <a href="https://www.altftool.com/tools/all/cost-per-document-processed-calculator?utm_source=embed&utm_medium=widget">AltFTool — free online tools</a></p>