AWS Textract Pricing 2026
Per-page rates by API, free tier limits, and a flat-rate alternative if you do not want to wire IAM, S3, and Lambda.
TL;DR
- AWS Textract charges per page and per API: $1.50/1,000 for plain text, $15/1,000 for tables, $50/1,000 for forms, $10/1,000 for receipts.
- Free tier covers 1,000 pages for the first 90 days of your AWS account, DetectDocumentText only — no free tier for tables, forms, or receipts.
- LensCopy is a flat-rate alternative: $0 with 10 pages lifetime free, $14.99/mo for 2,500 pages, with markdown tables and form-like structure included on every plan.
Bottom line: which should you use?
Bottom line: for individuals and teams who want extracted text without building an AWS pipeline, LensCopy is the better starting point. It runs in the browser, returns clean markdown with tables and equations on every plan, and bills a flat $14.99/month for 2,500 pages instead of metering $1.50–$50 per 1,000 pages across separate DetectDocumentText, AnalyzeDocument, and AnalyzeExpense APIs. There is no IAM, S3, or SDK to wire, and a permanent 10-page free tier to test on real documents. AWS Textract remains the right choice when your data already lives in S3, you process millions of pages a month, you need region-pinned compliance, or Textract is one node in a larger Bedrock or SageMaker workflow your engineers already own. For everyone else, LensCopy delivers usable output in under a minute.
LensCopy vs AWS Textract: feature comparison
Pricing snapshot (as of 2026-05-09)
LensCopy
- Free — $0 (10 pages lifetime)
- Basic — $14.99/mo (2,500 pages/month)
- Pro — $49.99/mo (10,000 pages/month)
- Business — Custom (Unlimited, dedicated host)
AWS Textract
- DetectDocumentText — ~$1.50 / 1,000 pages (Plain text only)
- AnalyzeDocument (Tables) — ~$15 / 1,000 pages (Adds table structure)
- AnalyzeDocument (Forms) — ~$50 / 1,000 pages (Adds key-value pairs)
- AnalyzeExpense — ~$10 / 1,000 pages (Receipts & invoices)
FAQ: AWS Textract Pricing 2026
Per-page rates depend on the API: DetectDocumentText (plain text) is $1.50 per 1,000 pages, AnalyzeDocument with Tables is $15 per 1,000 pages, AnalyzeDocument with Forms is $50 per 1,000 pages, and AnalyzeExpense (receipts and invoices) is $10 per 1,000 pages. AnalyzeID for IDs is $2.50 per 1,000 pages.
Yes, but it is limited. AWS offers a 90-day free tier covering the first 1,000 pages of DetectDocumentText only — tables, forms, receipts, and IDs are billed from page one. After 90 days, all usage is metered. LensCopy provides a permanent 10-page lifetime free tier with no expiration and no API restrictions.
DetectDocumentText returns bounding boxes around text. AnalyzeDocument additionally returns structured layout — table cells, form key-value pairs, or query answers — which is computed by a separate model. The 10x to 33x price jump reflects that extra structural inference, not just OCR.
For most use cases below 100K pages/month, yes. For AWS-native high-volume pipelines that already live in S3 + Lambda, Textract is purpose-built and remains the better fit.
Yes. LensCopy returns markdown tables ready to paste into a spreadsheet or doc. Textract returns a structured table block graph that you (or a library like textractor) emit as a final format.
Accuracy varies by document type. For printed business forms with clean layouts, both perform well. For documents containing equations, mixed handwriting, or non-standard layouts, LensCopy's vision-language model pipeline tends to reconstruct structure better.
Yes. Authenticated tiers expose extraction via API so the same engine that powers the web app powers your integrations.
Comparison data verified 2026-05-09. Loading interactive comparison…