Azure Document Intelligence vs Granite Docling
Pick Azure Document Intelligence for invoices, receipts and ID documents, especially if you already run on Azure and want a managed API with prebuilt models. Pick Granite Docling for English PDFs with tables, code and equations that you convert yourself at the cost of compute. The main split is a paid managed service versus a free, Apache-2.0, 258M-parameter model you host.
Azure Document Intelligenceazure/document-intelligence | Granite Doclingibm/granite-docling | |
|---|---|---|
| Type | OCR API | Open source |
| Price | 60 credits / page | 1 credit / page |
| Per 1,000 pages | $60.00 | $1.00 |
| Typical speed | 1.5 s / page | 1.5 s / page |
| Best for | Enterprise forms and IDs | Small-model document conversion |
| Inputs | PDF, images | PDF, images |
| Languages | 300+ languages | English, experimental CJK and Arabic |
| Max pages per request | 2,000 | 500 |
| Tables | ✓ Yes | ✓ Yes |
| Handwriting | ✓ Yes | No |
| Math | No | ✓ Yes |
| Bounding boxes | ✓ Yes | ✓ Yes |
| Confidence | ✓ Yes | No |
| Forms | ✓ Yes | No |
| Multilingual | ✓ Yes | No |
| On OCRRouter | Available | Available |
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 train anything for those document types.
- It runs as a managed service or as disconnected containers for on-prem use, and it fits teams already on Azure.
- Read OCR costs $1.50 per 1,000 pages, falling to $0.60 above 1M pages, and there is a free tier of 500 pages per month.
Choose Granite Docling if
- It is Apache-2.0 and free to run. You pay only for compute, and at 258M parameters it can run on CPU or a small GPU.
- It reports a table structure TEDS of 0.97 on FinTabNet, plus code recognition F1 of 0.988 and equation F1 of 0.968.
- It outputs DocTags, Markdown, HTML, LaTeX and bounding boxes, and it is built to work inside the Docling library.
Key differences
| Azure Document Intelligence | Granite Docling | |
|---|---|---|
| Licence | Proprietary | Apache-2.0 |
| How you run it | Managed Azure service, or containers including disconnected on-prem | Self-hosted only, typically inside Docling |
| Price | Read $1.50 per 1,000 pages; Layout and prebuilt $10 per 1,000; custom extraction $30 per 1,000 | Free; you pay for compute |
| Output formats | JSON, Markdown, plain text, bounding boxes, searchable PDF | DocTags, Markdown, HTML, LaTeX, bounding boxes |
| Table benchmark | 82.7% on RD-TableBench (independent) | TEDS 0.97 on FinTabNet (reported by the authors, a different test) |
| Languages | Not published | English primary; Japanese, Arabic and Chinese are experimental |
| Known failure | Stacked models multiply cost: Prebuilt plus Layout plus high-res add-on is $26 per 1,000 pages | Can loop on page headings and invoices; may print only exclamation marks on GPUs without bfloat16 such as T4 unless run in float32 |
Sources
- Azure pricing page
- Azure DI pricing breakdown
- Reducto RD-TableBench
- Granite-Docling model card
- HF discussion #24 repetition
What each one is
Azure Document Intelligence
Azure layout model with tables, selection marks and prebuilt invoice and ID readers.
Azure Document Intelligence detailsGranite Docling
Compact IBM vision model that converts pages to DocTags for the Docling pipeline.
Granite Docling detailsTry both on your own file
Run the file once with each model and compare the output.
Use one, fall back to the other
{"model": "azure/document-intelligence","fallbacks": ["ibm/granite-docling"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
Azure Document Intelligence alternativesGranite Docling alternativesBest OCR models in 2026