OCRRouter

Azure Document Intelligence vs docTR

Pick Azure Document Intelligence for invoices, receipts and ID documents when you want a managed service with prebuilt models, or when you already run on Azure. Pick docTR when you want free, Apache-2.0 word-level OCR with coordinates that you run and train yourself in Python. The main split is managed, per-page-priced extraction versus a self-hosted library that returns only words and lines.

Azure Document Intelligenceazure/document-intelligencedocTRmindee/doctr
TypeOCR APIOpen source
Price60 credits / page1 credit / page
Per 1,000 pages$60.00$1.00
Typical speed1.5 s / page0.9 s / page
Best forEnterprise forms and IDsWord-level OCR with positions
InputsPDF, imagesPDF, images
Languages300+ languagesMainly Latin scripts
Max pages per request2,000500
Tables✓ YesNo
Handwriting✓ YesNo
MathNoNo
Bounding boxes✓ Yes✓ Yes
Confidence✓ Yes✓ Yes
Forms✓ YesNo
Multilingual✓ YesNo
On OCRRouterAvailableAvailable

Prices are what you pay on OCRRouter. Speed is the provider's typical figure, not a benchmark we ran.

Which to choose

Choose Azure Document Intelligence if

  • It has prebuilt models for invoices, receipts, IDs, W-2 and contracts, so you do not have to build extraction yourself.
  • It outputs Markdown and searchable PDF as well as JSON, and it has a Layout model for tables (82.7% on RD-TableBench, an independent test).
  • It offers disconnected containers for on-prem use, and it fits enterprises already on Azure.

Choose docTR if

  • It is free under Apache-2.0, so you pay only for your own compute and there is no per-page fee.
  • It lets you pick detection and recognition architectures, with straight or rotated boxes, and train your own models.
  • It exports a nested page, block, line and word structure with coordinates as JSON, which suits custom extraction on invoices and receipts.

Key differences

Azure Document IntelligencedocTR
LicenceProprietaryApache-2.0
How you run itManaged Azure service, or containers including disconnected on-premPython library on PyTorch, with Docker images; CPU or NVIDIA GPU (GPU images need CUDA 12.2 or newer)
PriceRead $1.50 per 1,000 pages ($0.60 above 1M); Layout and prebuilt $10 per 1,000; custom extraction $30 per 1,000; 500 free pages a monthFree, self-hosted; you pay for compute
Output formatsJSON, Markdown, plain text, bounding boxes, searchable PDFJSON, plain text, bounding boxes
TablesLayout model; 82.7% on RD-TableBench, but costs $10 per 1,000 pages or moreNo table output; returns words and lines only
Prebuilt extractionInvoices, receipts, IDs, W-2, contractsNone prebuilt; a key information extraction predictor is included
Known issuesStacked models multiply cost (Prebuilt plus Layout plus high-res add-on is $26 per 1,000 pages); free tier covers only the first two pages of each documentSplit-crop merging misjoins readings across repeated characters (issue #2229); cropped images give poorer results (issue #2007); ONNX export of MASTER truncates decoding steps (issue #2224)

Sources

What each one is

Azure Document Intelligence

Azure layout model with tables, selection marks and prebuilt invoice and ID readers.

Azure Document Intelligence details

docTR

Open-source OCR library from Mindee with separate text detection and recognition models.

docTR details

Try both on your own file

Run the file once with each model and compare the output.

PDF, PNG, JPG, WebP, HEIC. Up to 20 MB each, 10 files.No file?

Use one, fall back to the other

{"model": "azure/document-intelligence","fallbacks": ["mindee/doctr"],"file_url": "https://example.com/document.pdf"}

Questions

More comparisons

Azure Document Intelligence alternativesdocTR alternativesBest OCR models in 2026