“We were scaling document volume by nearly 50 percent every month —
Nutrient handled it without us adding resources.”
Harvey
compare
Amazon Textract vs. Nutrient
Looking for an Amazon Textract alternative? Nutrient’s Data Extraction API covers the same OCR, table, form, and field extraction, and ties every extracted field back to its place on the page. Nutrient also offers the viewers, editors, forms, and signing the same documents need. Here’s where Nutrient is stronger, where Textract still leads, and what switching involves.
At a glance
| Nutrient | Amazon Textract | |
|---|---|---|
| Core approach | Four processing modes, from fast text extraction to vision
language models. Every extracted value is checked against the
source text. | An AWS service with separate APIs for OCR, forms and tables, and
invoices, IDs, and loan files. Output is a block graph your code
reassembles. |
| Scope | ||
| Deployment |
Lufthansa
Disney
Autodesk
UBS
Dropbox
IBM
Lufthansa
Disney
Autodesk
UBS
Dropbox
IBM
Textract fits pipelines built inside AWS. Nutrient fits wherever your product runs.
Nutrient
Nutrient is for teams that embed extraction in their own product, on any cloud or on their own hardware. Each call returns Markdown or typed JSON that an LLM can use directly, with every field tied to its place on the page. The same vendor covers viewing, editing, forms, and signing.
Amazon Textract
Textract is strongest when the whole pipeline already lives in AWS. Documents sit in Amazon S3, IAM policies control access, and asynchronous jobs report back through Amazon SNS. Its specialized APIs read invoices, US identity documents, and mortgage packages without a schema to write.
Nutrient vs. Amazon Textract on ParseBench
ParseBench is LlamaIndex’s public document parsing benchmark, and AWS Textract is one of the entries on its leaderboard. Here’s how Nutrient’s agentic mode compares, dimension by dimension. Tables are Textract’s strongest dimension.
| ParseBench (0–100) | Nutrient (agentic) | AWS Textract | Higher score |
|---|---|---|---|
| Overall | 76.4 | 47.9 | Nutrient |
| Charts | 83.5 | 6.0 | Nutrient |
| Visual grounding | 77.5 | 70.4 | Nutrient |
| Semantic formatting | 64.7 | 3.7 | Nutrient |
| Content faithfulness | 84.5 | 74.8 | Nutrient |
| Tables | 72.1 | 84.6 | Amazon Textract |
Most of the overall gap comes from charts and semantic formatting. Semantic formatting scores formatting that carries meaning, such as strikethrough, superscript and subscript, bold, and title hierarchy, which Textract’s block output doesn’t preserve. On content faithfulness, agentic mode leads by 9.7 points; on tables, Textract leads by 12.5 points. Nutrient’s scores are its row on the public ParseBench leaderboard, Nutrient DWS (Agentic), which runs agentic mode at 18 credits per page. Nutrient submitted it using ParseBench’s own scoring against the live API. In our internal test of the latest engine on 14 October 2026, which includes improvements that haven’t reached the live API yet, agentic mode scored 82.8 overall and understand mode, at 9 credits, scored 64.0. Internal test scores aren’t independently verified. AWS Textract: parsebench.ai, retrieved 15 October 2026. Gaps under 1 point are shown as level. Every mode and competitor is on the Data Extraction API benchmarks page.
Nutrient vs. Amazon Textract, feature by feature
Capability by capability, including where Textract wins.
| Nutrient | Amazon Textract | Winner | |
|---|---|---|---|
| Self-hosting | The Java and Python SDKs, sold separately, run extraction on your
servers, even offline. Schema extraction needs a model server,
local or hosted. | AWS Regions only. No on-premises or offline option. | Nutrient |
| Languages | OCR in 100+ languages, including Chinese, Japanese, Korean,
Arabic, and Cyrillic. | Six European languages. Handwriting, invoices, IDs, and Queries
are English-only. | Nutrient |
| LLM-ready output | Markdown, or page elements in reading order, from one call. | A block graph of lines, words, and cells that your code
reassembles. | Nutrient |
| Excel and spreadsheet input | Reads XLSX workbooks cell by cell and returns tables,
headings, and notes in the same output as PDFs. | Not supported. Input formats are JPEG, PNG, PDF, and TIFF. | Nutrient |
| Grounding checks | Every field is labeled with how well it matched the source text. | Values are read straight off the page, with confidence and boxes
per block. | Different approaches |
| Review inside your own product | The Nutrient viewer, sold separately, shows each value on the
page in your app. | Amazon A2I review loops are closed to new customers. You build
the review screen. | Nutrient |
| Document classification | Send your labels with the request. No training needed. | No classification API. AnalyzeLending sorts mortgage pages only. | Nutrient |
| Full document platform | Viewing, editing, redaction, signing, comparison, and conversion. | Text detection and document analysis only. | Nutrient |
| Targeted field extraction | JSON shaped to your schema, with a citation for every field. | Queries answer up to 15 questions a page, tunable with Custom
Queries. | Different approaches |
| Handwriting | Printed-style in understand mode, cursive in agentic mode. | Handwriting and print in the same call. | Draw |
| Native AWS integration | A REST API with API keys, from any cloud or your own servers. | IAM access control, S3 input and output, and GovCloud. | Amazon Textract |
| Prebuilt APIs for invoices, IDs, and lending | One schema-driven endpoint. You define the fields for each
document type. | Built-in fields for invoices, US IDs, and loan files, with no
schema to write. | Amazon Textract |
| Signature detection | The Java and Python SDKs detect signature fields and whether
they’re signed. Not in the cloud API. | Detects signatures anywhere on the page. | Amazon Textract |
| Async batch jobs | Async parse, extract, and classify: submit a run and poll for the
result, with up to 15 minutes of processing per run. | Async jobs from S3 with SNS alerts, up to 3,000 pages per file. | Amazon Textract |
What only Nutrient offers beyond extraction
Textract stops at text detection and document analysis. Nutrient is a full document platform, so the same vendor also powers the human-facing document work your application needs. That means one vendor instead of two; the SDKs are sold separately.
| Nutrient | Amazon Textract | |
|---|---|---|
| Document viewer and rendering | ||
| Annotations and markup | ||
| Form filling and editing | ||
| Redaction | ||
| eSignatures | ||
| Document comparison | ||
| Conversion and generation | ||
| Mobile SDKs |
“We don’t think any other tools have the breadth and the ease of use
that Nutrient has. We certainly have evaluated other companies over the
years. And every time we do that, we’ve come back to Nutrient.”
GOVENDA
How each vendor bills
Textract prices each API and feature separately. Nutrient’s rate depends on the processing mode.
| Nutrient | Amazon Textract | |
|---|---|---|
| Free usage | 5,000 credits a month, about 277 to 3,333 parse pages in auto, depending on the mode selected for each page | Three months for new AWS customers, 100 to 2,000 pages a month by
API |
| Text detection (OCR) | $1.50 per 1,000 pages (structure mode) | $1.50 per 1,000 pages |
| Layout, tables, and forms | $9 per 1,000 pages, all three in
one call | $4 (Layout), $15 (Tables), or $50 (Forms) per 1,000 pages |
| Queries and schema extraction | $7.50 to $24 per 1,000 pages in auto; $15 in understand; $7 to $24 across explicit modes | $15 to $70 per 1,000 pages, by feature mix |
| Invoices, IDs, and loan files | Schema extraction, priced as above | $10 to $70 per 1,000 pages, by API |
| Classification | $1 per 1,000 pages | No classification API |
Nutrient prices use the Pro plan, $500 a month for 500,000 credits (Starter costs 18% more). Extraction includes parsing. Auto routes each page to structure, understand, or agentic and bills at that mode’s rate. The lowest explicit-mode rates use text mode, which skips OCR and is not selected by auto. Textract prices are for US East (N. Virginia) and the first 1 million pages a month; they drop above that. Prices cover extraction; the SDKs are sold separately. Full rates: Nutrient pricing and Textract pricing (checked 23 September 2026).
Nutrient Extract Studio for macOS
Test Nutrient on your own documents
Extract Studio is a desktop app that runs Nutrient’s extraction engine on your Mac. With an AI model running locally through Ollama or LM Studio, your documents stay on the machine while you compare results. It runs on Apple silicon Macs. You need a Nutrient account, and each run uses Data Extraction credits.
What moving from Textract involves
The Data Extraction API maps Textract’s text detection and document analysis calls onto three REST endpoints. Most of the work is mapping calls, writing schemas for the fields you read today, and replacing block-graph parsing code.
| Amazon Textract | Nutrient |
|---|---|
| DetectDocumentText | POST /extraction/parse in structure mode runs OCR and
returns page elements with coordinates, or Markdown. |
| AnalyzeDocument: Layout, Tables, Forms | POST /extraction/parse in understand mode returns layout
and tables. Use the extract endpoint with a JSON Schema for named fields.
Start with auto when you want the processing mode selected per page. |
| Queries and Custom Queries | POST /extraction/extract returns JSON shaped to your schema.
Describe each field the way you’d phrase the question. |
| AnalyzeExpense and AnalyzeID | POST /extraction/extract with your own invoice, receipt,
or ID schema. Plans with data retention don’t allow government-ID data,
so check your plan first. |
| AnalyzeLending | Classify each document, then extract from it. To split a loan
package first, use the Java and Python SDKs. |
| Signatures | Not in the cloud API. The Java and Python SDKs detect signature
fields and whether they’re signed. |
| Asynchronous jobs (Start and Get) | Async parse, extract, and classify routes: submit a run and poll
for the result, like Start and Get. |
| Block objects | Page elements in reading order, Markdown, or schema JSON. No
graph to traverse. |
| Amazon Augmented AI review loops | Route fields by match label in your code, and show each value on
the page with the Nutrient viewer. |
The extract endpoint accepts a subset of JSON Schema: an object root with type, properties, required, items, description, enum, and date formats. Flatten schemas that use $ref, anyOf, oneOf, or validation ranges before you send them. Limits are 128 KB, 2,000 schema nodes, 200 properties per object, 50 enum values per field, and five levels of nesting. See define a schema.
Four steps to switch
Get an API key
Sign up and get 5,000 free credits every month, with no credit card required.
Run both on the same documents
Send a representative sample through both APIs. Start with auto for per-page routing and measure quality and usage. Choose agentic explicitly to compare the mode benchmarked above; understand mode costs half the credits and covers layout, tables, and forms.
Compare fields and citations
Check extracted values against Textract’s, and route fields marked fuzzy_match or not_found to review.
Move production traffic
Move document types over one at a time, once each meets your accuracy bar.
Enterprise switch offer
Start on the free plan and test your own documents with 5,000 credits a month. When you’re ready to move a production workload from Amazon Textract, talk to us: Enterprise teams that sign a Data Extraction agreement get 3,000,000 additional credits, about six months of Pro plan usage, to cover the migration.
Why teams choose Nutrient as an Amazon Textract alternative
Process outside AWS
Use the cloud API from any cloud, or run the Java and Python SDKs, sold separately, on your own servers, even offline.
One platform, not just extraction
Parse and extract. Then view, edit, fill forms, redact, sign, compare, and convert with the same vendor. Textract stops at text detection and document analysis.
100+ languages
OCR in 100+ languages, including Chinese, Japanese, Korean, Arabic, and Cyrillic scripts.
Output you can check
Markdown or typed JSON an LLM can use directly. Every extracted field gets a match label, plus its source region and a confidence signal when the value is found on the page.
Start free
5,000 Data Extraction API credits every month, no credit card required. Pick the cheapest mode that meets your accuracy bar.
Frequently asked questions
What is Amazon Textract?
Amazon Textract is an AWS machine learning service that detects printed text and handwriting and extracts structured data from scanned documents. DetectDocumentText runs OCR. AnalyzeDocument adds Forms, Tables, Queries, Custom Queries, Signatures, and Layout. AnalyzeExpense, AnalyzeID, and AnalyzeLending cover invoices and receipts, US identity documents, and mortgage packages. It runs only in AWS Regions and returns a graph of block objects.
Is Nutrient a good Amazon Textract alternative?
Yes, Nutrient is a good Amazon Textract alternative for teams that need to extract outside AWS, read documents in more than six languages, or feed an LLM directly. Nutrient offers a cloud API and self-hosted Java and Python SDKs, OCR in 100+ languages, and Markdown and schema JSON with every field tied back to the page. On ParseBench, Nutrient’s agentic mode scores higher than AWS Textract overall and on charts, visual grounding, semantic formatting, and content faithfulness. AWS Textract scores higher on tables. Textract has the edge on native AWS integration, prebuilt APIs for invoices, US IDs, and mortgage packages, signature detection, and asynchronous batch jobs on S3.
What are the best alternatives to Amazon Textract?
The main alternatives are Nutrient, Azure Document Intelligence, Google Document AI, LlamaParse, Reducto, Unstructured, and Docling. Nutrient stands out as a full document platform that runs in the cloud or on your own hardware, with every extracted field tied back to its source on the page.
Amazon Textract vs. Nutrient for RAG — which is better?
Nutrient returns reading-ordered Markdown and schema-shaped JSON with per-field citations, which a retrieval-augmented generation (RAG) pipeline can chunk and an LLM can use directly. Textract returns a block graph that your code must traverse and linearize. On ParseBench semantic formatting, AWS Textract scores 3.7, against 64.7 for Nutrient’s agentic mode. Textract scores higher on ParseBench tables, so it stays a strong choice for table-heavy documents in an existing AWS pipeline.
Can I run document extraction on-premises or outside AWS?
Yes, with Nutrient. The Nutrient Java and Python SDKs run extraction on your own servers, even offline, under an SDK license. The cloud API can be called from any cloud. Amazon Textract runs only in AWS Regions, including AWS GovCloud (US), with no on-premises, edge, or offline option.
To test Nutrient on your own documents before deciding, use Nutrient Extract Studio for macOS, a desktop app that runs the same extraction engine on an Apple silicon Mac. With an AI model running on that Mac, your documents stay on it. The app needs a Nutrient account and an internet connection to sign in, and runs use Data Extraction credits.
Does Nutrient support non-English documents?
Yes, Nutrient performs OCR in 100+ languages, including Chinese, Japanese, Korean, Arabic, and Cyrillic scripts. Amazon Textract extracts printed text, forms, and tables in six languages: English, French, German, Italian, Portuguese, and Spanish. Its handwriting recognition, invoice and receipt analysis, ID analysis, and Queries are English-only.
How does pricing compare to Amazon Textract?
Textract charges per 1,000 pages for the first 1 million pages a month in US East: $1.50 for DetectDocumentText, $15 for Tables, $50 for Forms, and $70 for Tables + Forms + Queries. Auto is Nutrient’s recommended starting mode: it routes each page to structure, understand, or agentic and bills at that mode’s rate. Nutrient bills credits per page: 1 for explicit text parsing, 1.5 for structure, 9 for understand, and 18 for agentic, with schema extraction adding 6 credits per page and 5,000 free credits every month. At the Pro plan rate of $0.001 per credit ($500 a month for 500,000 credits), auto costs $1.50 to $18 per 1,000 pages for parsing and $7.50 to $24 for schema extraction. Across explicit modes, the ranges start at $1 and $7 respectively with text mode, which skips OCR. On the Starter plan ($59 a month for 50,000 credits), every Nutrient figure is 18% higher. Plan prices and pay-as-you-go rates are on the pricing page. Talk to our team for a comparison scoped to your document mix.
How do I move from Amazon Textract to Nutrient?
Map DetectDocumentText and AnalyzeDocument to Nutrient’s /extraction/parse, and Queries, AnalyzeExpense, and AnalyzeID to /extraction/extract with a JSON Schema. Replace block-graph parsing with Markdown or typed elements, and run a sample through both APIs on the free plan. When you’re ready to move a production workload, talk to us about the enterprise switch offer: 3,000,000 additional credits for teams that sign a Data Extraction agreement.
Can Nutrient do more than extraction?
Yes, Nutrient’s Data Extraction API is the parsing layer of a full document platform. Connect its output to Nutrient’s Document Web Services (DWS) APIs for conversion, redaction, generation, and signing, and to Nutrient SDKs when people need to review, edit, annotate, fill, or sign documents in your application. Textract stops at text detection and document analysis.
EXPLORE
Keep comparing
Reducto
Reducto is a strong agentic extraction platform, with webhooks and batch processing in its cloud API. Nutrient is the broader document platform: a match label on every extracted field, plus viewing, editing, signing, and conversion from the same vendor.
Unstructured.io
Unstructured.io is a strong RAG-ingestion platform with partitioning, chunking, schema extraction, and 40+ connectors. Nutrient adds a match label on every extracted field, plus viewing, editing, signing, and conversion from the same vendor.
LlamaIndex
LlamaParse scores higher than Nutrient on ParseBench, LlamaIndex’s own benchmark. Nutrient adds a match label on every extracted field, a viewer to check each citation on the page, and viewing, editing, and signing from the same vendor.
Documents in, structured data out
5,000 free Data Extraction API credits per month, with no credit card required. Parse and extract source-grounded data your AI workflows can trust, in the cloud or on your own infrastructure.