AWS Textract Pricing 2026: Per-Page Rates by Feature, the Free Tier, and What a Real Intake Workload Costs

As of October 24, 2026, Amazon Textract charges per page by feature: $0.0015 for plain text, $0.01 for invoices and receipts, $0.015 for tables or queries, $0.025 for IDs and $0.05 for forms, adding up to $0.07 for forms, tables and queries together. New AWS customers get 3 free months. A property manager screening 300 rental applications a month pays about $35 with Queries, or about $125 to $134 with forms switched on.

JM
Justin McKelvey
October 24, 2026

How much does AWS Textract cost? The short answer

As of October 24, 2026, Amazon Textract charges per page, and the rate depends on which feature reads it. Plain text extraction is $0.0015 a page ($1.50 per 1,000). Invoices and receipts are $0.01, tables $0.015, Queries $0.015, IDs $0.025, and forms $0.05. Combine features and the rates add up, to $0.07 a page for forms, tables and queries together. New AWS customers get a 3-month free tier. For a property manager screening 300 rental applications a month, the bill runs from about $35 to $134 depending only on which features the build calls.

Every rate on this page comes from Amazon's own Textract pricing page, read October 24, 2026, where Amazon states its per-page prices through worked examples for the US West (Oregon) region. Prices vary by region, so check the live page for yours.

Textract prices by feature

Textract has five APIs. Each bills per page, with a lower rate for pages past 1 million in a month.

API and featureFirst 1 million pages a monthOver 1 million
Detect Document Text (text and handwriting)$0.0015 a page$0.0006
Analyze Document, Tables (layout included free)$0.015$0.010
Analyze Document, Forms (key-value pairs)$0.05$0.04
Analyze Document, Queries$0.015Not shown in the examples
Analyze Document, Tables + Queries$0.020$0.015
Analyze Document, Forms + Tables + Queries$0.070$0.055
Analyze Document, Signatures alone$0.0035$0.0014
Analyze Document, Custom Queries$0.025$0.015
Pretrained Forms + Custom Queries$0.065$0.050
Analyze Expense (invoices and receipts)$0.01$0.008
Analyze ID (driver's licenses, passports)$0.025 for the first 100,000$0.01 after 100,000
Analyze Lending (mortgage packets)$0.07 per classified and extracted page$0.055

Three things on Amazon's page change the math. Layout is free with any combination of forms, tables and queries. Signature detection is free when combined with any of those features. And training the adapters behind Custom Queries is free. Analyze Expense and Analyze ID include the OCR text in their output, so you do not pay Detect Document Text on top.

Free decision sheet

The AP Automation Decision Sheet

Baseline your per-invoice cost, match features to your AP complexity, and get a defensible SaaS-vs-custom call in 20 minutes.

Also subscribes you to The Service Business AI Brief. No spam, unsubscribe anytime.

Is AWS Textract free?

Only for three months, and only for new AWS customers. Amazon's free tier covers, per month: 1,000 pages of Detect Document Text; 1,000 pages of Analyze Document using Signatures alone, or 100 pages using Forms, Tables and Layout; 100 pages each for Queries and its combinations; 100 pages of Analyze Expense; 100 of Analyze ID; and 2,000 pages of Analyze Lending. Custom Queries has no free tier. That is enough to test a document type, not to run an office on.

A worked bill: a property manager screening rental applications

Amazon's examples run from 5,000 pages to 5 million, which is not a small business. Here is one that is. A property management company with about 600 doors screens 300 rental applications a month. Each packet is a 4-page application form, 2 single-page pay stubs and a driver's license, so 1,200 application pages, 600 pay stub pages and 300 IDs. The build needs about 10 fields from each page: name, employer, income, move-in date, prior address and so on.

How the build reads the packetsMathMonthly cost
Route A: Queries. Ask for the 10 fields by name on every page1,800 pages x $0.015 + 300 IDs x $0.025about $34.50
Route B: Forms + Tables. Pull every key-value pair and table, then pick fields out1,800 x $0.065 + 300 x $0.025about $124.50
Route C: Forms + Tables + Queries. Everything, to be safe1,800 x $0.070 + 300 x $0.025about $133.50

Same documents, same answers needed, and a 3.9x spread. The Forms feature is the expensive line at $0.05 a page, more than three times Queries, and a build that switches it on "to be safe" pays for every key-value pair on a page when it needs ten fields. The signature check on the application is free in all three routes, because it rides along with the other features.

Route A has a limit worth knowing: Queries work best on fields that are printed on the page with a label. If the application forms vary a lot between owners or have handwritten sections, a Custom Queries adapter at $0.025 a page (about $45 for these 1,800 pages, plus $7.50 for IDs, so about $52.50) is the middle road. Our guide to AI for property management covers where screening sits next to leasing and maintenance.

The same workload on Azure Document Intelligence

Azure prices below are from Microsoft's Document Intelligence pricing page, read October 24, 2026, and our Azure Document Intelligence pricing breakdown runs a bookkeeping workload through it. Azure has no prebuilt pay-stub or rental-application model on its list, so the applications and pay stubs go through custom extraction at $30 per 1,000 pages (1,800 pages, $54) and the IDs through its prebuilt ID model at $10 per 1,000 ($3). That is about $57 a month: cheaper than Textract's Forms routes, dearer than Queries. On invoices and receipts the two services tie at $10 per 1,000 pages, and on IDs Azure is cheaper.

The lesson carries across both clouds: the feature list, not the vendor, sets the bill. Ask whoever builds it which features each document type calls, and why.

What the Textract price page leaves out

  • The rest of AWS. Documents usually land in S3, and code (often a Lambda function) calls Textract and writes the results somewhere. Those services bill separately.
  • Human review. Textract returns each field with a confidence score. What happens to a 70%-confident income figure is a workflow decision, and any review step is outside these prices.
  • Getting data into your software. Textract hands back JSON. Moving it into a screening tool, AppFolio, Buildium or a spreadsheet is the integration work, and it is most of the build.
  • Region. Amazon's examples are for US West (Oregon). Its price table loads by region, so a different region can cost more.

Is AWS Textract worth it?

For a business that already runs on AWS and has a builder, yes. At $34.50 a month for 300 applications, extraction costs about 12 cents a packet, far less than the minutes someone spends keying it. It is cheaper than the per-document tools in our Docsumo vs Rossum vs Nanonets comparison, but those include review screens and integrations that Textract leaves to you.

It is not worth it as a do-it-yourself project for an office without a developer. Textract is an API with no screen for your team to use. If nobody will maintain the code, buy a tool. Our guide on how to automate data entry with AI covers when each path makes sense.

Who can build a Textract document workflow for my business?

A developer who knows AWS can get one document type extracting in a few days. The work that takes longer is the rest: sorting the packets, choosing the cheapest features per document, handling low-confidence fields and writing the results into the system your team actually uses. SuperDupr, the AI ops shop Justin McKelvey runs, builds those pipelines. Our published price for AI document processing is $8,000 to $20,000. Most AI builds go live in 2 to 4 weeks, support afterwards runs $200 to $2,000 a month depending on volume, and the business owns what we build. Textract usage stays billed to your own AWS account at Amazon's rates; we do not resell it. The first step is a free 30-minute strategy session, booked from that page.

Sources

  • Amazon Textract pricing, aws.amazon.com/textract/pricing, read October 24, 2026: the five APIs, free tier (3 months, new customers, pages per month by feature, no free tier for Custom Queries), free layout and signatures in combinations, free adapter training, OCR included in Analyze Expense and Analyze ID, and per-page rates as stated in Amazon's 18 pricing examples for US West (Oregon).
  • Azure Document Intelligence in Foundry Tools pricing, azure.microsoft.com/en-us/pricing/details/ai-document-intelligence, US dollars, read October 24, 2026: prebuilt model list and $10 per 1,000 pages, custom extraction $30 per 1,000 pages.
  • Worked bill and Azure comparison: SuperDupr's arithmetic on the published rates above, for an assumed 300 applications a month.

Frequently Asked Questions

Buyer's Guide Pricing Automation

AI Document Processing $8K–$20K

We build this for service businesses — transparent pricing, fixed scope.

See how it works →

Free decision sheet

The AP Automation Decision Sheet

Baseline your per-invoice cost, match features to your AP complexity, and get a defensible SaaS-vs-custom call in 20 minutes.

Also subscribes you to The Service Business AI Brief. No spam, unsubscribe anytime.

The Service Business AI Brief

One practical AI win for service businesses, every other week. No fluff, unsubscribe anytime.

Related Articles

Pricing

Azure Document Intelligence Pricing 2026: Per-Page Rates, the Free Tier, Commitment Tiers and a Real Monthly Bill

As of October 24, 2026, Azure Document Intelligence costs $1.50 per 1,000 pages to read text, $10 per 1,000 for any prebuilt model (invoices, receipts, IDs, layout), $30 per 1,000 for custom extraction and $3 for custom classification, with 500 free pages a month. A 25-client bookkeeping practice sending about 5,000 pages a month pays roughly $70, and the smallest commitment tier only pays off past 19,000 prebuilt pages.

Pricing

Telnyx Pricing 2026: Voice, SMS, Numbers and Voice AI Rates, and What a Small Business Really Pays

As of October 23, 2026, Telnyx is pay as you go: a US local number is $1 a month ($0.10 more for texting), a call costs a $0.002-a-minute API fee plus SIP trunking from $0.0032 a minute in and $0.005 out, and a text is $0.004 per part plus carrier fees. Its voice AI engine is $0.05 a minute with the model billed per token. For a 400-call month at a small service business, the phone side runs about $14.50, against about $22.50 on Twilio.

Pricing

Mapbox Pricing 2026: Free Tier, Map Loads, Geocoding, and Mapbox vs Google Maps on Real Bills

Mapbox is free up to 50,000 web map loads, 100,000 geocoding requests and 100,000 routing requests or matrix elements a month, as of October 23, 2026, then $5 per 1,000 map loads and $2 per 1,000 routing calls. Address Autofill is the expensive line at $12.50 per 1,000 sessions. Here are the same three workloads priced on Mapbox and Google Maps, and the geocoding rule that matters for a CRM.

Ready to Implement AI in Your Business?

Book a free strategy session to see how the concepts in this article can work for your specific business.