Donut vs Google Cloud Vision OCR
Pick Google Cloud Vision OCR when you need plain text, bounding boxes or handwriting from photos and scanned pages at high volume, with no training step. Pick Donut when you have a narrow, fixed-layout document type such as receipts or forms and want structured JSON from a self-hosted model you fine-tune yourself. The main split is a ready-to-call hosted API versus a base model that you must train and run.
Donutnaver/donut | Google Cloud Vision OCRgoogle/cloud-vision | |
|---|---|---|
| Type | Open source | OCR API |
| Price | 1 credit / page | 2 credits / page |
| Per 1,000 pages | $1.00 | $2.00 |
| Typical speed | 1.0 s / page | 1.0 s / page |
| Best for | Receipt and form fields after fine-tuning | Photos and simple scans |
| Inputs | images | PDF, images |
| Languages | English, Chinese, Japanese, Korean | 100+ languages |
| Max pages per request | 100 | 2,000 |
| Tables | No | No |
| Handwriting | No | ✓ Yes |
| Math | No | No |
| Bounding boxes | No | ✓ Yes |
| Confidence | No | ✓ Yes |
| Forms | ✓ Yes | No |
| Multilingual | No | ✓ Yes |
| 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 Donut if
- You can self-host it under the MIT licence and pay only for compute, with no per-page fee.
- It maps a document image straight to structured JSON with no separate OCR step, which suits receipt and form field extraction after fine-tuning.
- It also supports document classification, and a practitioner on Hacker News reports that about 200 documents of fine-tuning can reach human-level accuracy on narrow data.
Choose Google Cloud Vision OCR if
- It works without training: you call TEXT_DETECTION for photos or DOCUMENT_TEXT_DETECTION for dense documents over REST or client libraries.
- It returns a page, block, paragraph, word and symbol hierarchy with bounding polygons, and supports handwriting through DOCUMENT_TEXT_DETECTION and language hints.
- It scales for volume: the first 1,000 units per month are free, the price drops to $0.60 per 1,000 above 5M units, and async batch takes up to 2,000 image files per request.
Key differences
| Donut | Google Cloud Vision OCR | |
|---|---|---|
| How you run it | Self-hosted base model, available in Hugging Face Transformers | Hosted API on Google Cloud, no on-premise option |
| Licence | MIT | Proprietary |
| Price | Free, self-hosted (you pay for compute) | First 1,000 units per month free, then $1.50 per 1,000 up to 5M, $0.60 per 1,000 above. Each PDF page is one unit. |
| Ready out of the box | No. The base checkpoint is pretrain-only and must be fine-tuned before use. | Yes. Billing must be enabled, even for the free 1,000 units. |
| Output formats | JSON, plain text | JSON, plain text, bounding boxes. No Markdown, so you rebuild reading order yourself. |
| Tables and layout | Extracts fields from narrow, fixed-layout documents after fine-tuning. Not suited to full-page Markdown conversion. | No table or layout structure. Google points to Document AI for forms and tables. |
| Known failure | Misses tiny text in large images because of input resolution limits. A Hacker News user found results on personal documents did not match the promised scores. Transformers v5 dropped image-to-text pipeline support, so you need direct loading or v4. | PDF and TIFF require async processing via Cloud Storage. |
Sources
What each one is
Donut
OCR-free document understanding model. Fine-tuned for receipts, forms and document Q&A.
Donut detailsGoogle Cloud Vision OCR
Google Cloud's image OCR with text and document text detection, word boxes and confidence.
Google Cloud Vision OCR 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": "naver/donut","fallbacks": ["google/cloud-vision"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
Donut alternativesGoogle Cloud Vision OCR alternativesBest OCR models in 2026