OCRRouter

Azure Document Intelligence vs SmolDocling

Pick Azure Document Intelligence for invoices, receipts and IDs in production, where its prebuilt models and managed service matter more than per-page cost. Pick SmolDocling only for English experiments on your own hardware. It is a free preview model that the authors have replaced with Granite-Docling.

Azure Document Intelligenceazure/document-intelligenceSmolDoclinghuggingface/smoldocling
TypeOCR APIOpen source
Price60 credits / page1 credit / page
Per 1,000 pages$60.00$1.00
Typical speed1.5 s / page1.0 s / page
Best forEnterprise forms and IDsVery cheap page conversion
InputsPDF, imagesPDF, images
Languages300+ languagesMainly English
Max pages per request2,000500
Tables✓ Yes✓ Yes
Handwriting✓ YesNo
MathNo✓ Yes
Bounding boxes✓ Yes✓ Yes
Confidence✓ YesNo
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 ships prebuilt models for invoices, receipts, IDs, W-2 and contracts, so you do not have to build field extraction yourself.
  • It has a stated 82.7% on RD-TableBench table accuracy from an independent test, and a managed service with disconnected containers for on-prem use.
  • Read OCR costs $1.50 per 1,000 pages ($0.60 above 1M), with 500 free pages per month.

Choose SmolDocling if

  • It is free to use under a permissive licence, and you only pay for compute.
  • At 256M parameters it runs on modest hardware, at about 0.35 seconds per page on an A100 with vLLM.
  • One model outputs DocTags, Markdown, HTML, JSON and LaTeX, and handles text, tables, code and formulas with location tags.

Key differences

Azure Document IntelligenceSmolDocling
Release statusGenerally available Microsoft servicePreview release; authors point users to Granite-Docling
How you run itManaged API, or containers (including disconnected)Self-host only
LicenceProprietaryApache-2.0 per the model summary (metadata tag says CDLA-Permissive-2.0)
PriceRead $1.50 per 1,000 pages; Layout and prebuilt $10 per 1,000; custom extraction $30 per 1,000Free, you pay for compute
Table accuracy82.7% on RD-TableBench (independent)0.82 TEDS structure on FinTabNet 150dpi (reported by the authors, a different metric)
LanguagesNot publishedEnglish only
Known failureStacked models multiply cost: Prebuilt plus Layout plus high-res add-on is $26 per 1,000 pages. The free tier covers only the first two pages of each document.Output can loop on some images and become very slow (Docling issue #1515); endless token repetition and missing closing tags are listed as typical defects

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

SmolDocling

Tiny 256M vision model from Hugging Face and IBM that outputs DocTags. Predecessor of Granite Docling.

SmolDocling 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": ["huggingface/smoldocling"],"file_url": "https://example.com/document.pdf"}

Questions

More comparisons

Azure Document Intelligence alternativesSmolDocling alternativesBest OCR models in 2026