PaddleOCR-VL vs dots.ocr
Pick PaddleOCR-VL for multilingual scans, academic papers and reports on modest GPUs, since it is under 1B parameters and covers 109 languages under Apache-2.0. Pick dots.ocr when you need layout boxes, HTML tables, LaTeX formulas or SVG charts from one model. Avoid dots.ocr on forms with long dot leaders and on very large dense tables.
PaddleOCR-VLpaddle/paddleocr-vl | dots.ocrrednote/dots-ocr | |
|---|---|---|
| Type | Open source | Open source |
| Price | 6 credits / page | 2 credits / page |
| Per 1,000 pages | $6.00 | $2.00 |
| Typical speed | 1.3 s / page | 1.5 s / page |
| Best for | CJK documents | Multilingual document parsing |
| Inputs | PDF, images | PDF, images |
| Languages | 109 languages | 100+ languages |
| Max pages per request | 500 | 500 |
| Tables | ✓ Yes | ✓ Yes |
| Handwriting | No | No |
| Math | ✓ Yes | ✓ Yes |
| Bounding boxes | ✓ Yes | ✓ Yes |
| Confidence | ✓ Yes | No |
| Forms | No | No |
| Multilingual | ✓ Yes | ✓ 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 PaddleOCR-VL if
- It has 0.9B parameters, so it runs on much smaller GPUs than 3B or 7B document models.
- It supports 109 languages in one model, including Cyrillic, Arabic, Devanagari and Thai scripts.
- Version 1.6 scores 96.33 on OmniDocBench v1.6, the top figure its authors report for that benchmark.
Choose dots.ocr if
- It returns layout boxes, categories and text in one pass, with tables as HTML and formulas as LaTeX.
- It offers more output formats: JSON, Markdown, HTML, LaTeX, bounding boxes and SVG.
- The dots.mocr (v1.5) release scores 83.9 on olmOCR-bench, up from 79.1 for the original, and the dots.mocr-svg variant converts charts and diagrams to SVG.
Key differences
| PaddleOCR-VL | dots.ocr | |
|---|---|---|
| Licence | Apache-2.0 | MIT, plus a separate dots.ocr license agreement file in the repository |
| Model size | 0.9B parameters | 3B parameters (1.7B LLM backbone) |
| Output formats | Markdown and JSON | JSON, Markdown, HTML, LaTeX, bounding boxes and SVG |
| olmOCR-Bench | 80.0 overall (independent result) | 79.1 for dots.ocr and 83.9 for dots.mocr (author-reported) |
| Languages | 109 languages, including Cyrillic, Arabic, Devanagari and Thai | Described as multilingual; no language count published |
| Hardware | NVIDIA GPU recommended, CPU possible but slow; no minimum VRAM stated | Single CUDA GPU with vLLM is the documented path; no VRAM figure stated |
| Known failure | Version 1.6 decodes about 11x slower because use_cache ships as false, hangs on blurry images, and regresses Russian and Kazakh Cyrillic on born-digital PDFs | Runs of ellipses or underscores can make output repeat endlessly, and a table above 1000 grid cells raises an error and drops the page |
Sources
- Hugging Face model card
- PaddleOCR-VL-1.6 model card
- Issue #18453 use_cache slowdown
- dots.ocr GitHub README
- dots.ocr blog / limitations
- Docling issue #5146
What each one is
PaddleOCR-VL
Open-source document model from Baidu. Excellent on Chinese, Japanese and Korean.
PaddleOCR-VL detailsdots.ocr
Single vision model from rednote that does layout detection and content recognition together.
dots.ocr detailsTry 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": "paddle/paddleocr-vl","fallbacks": ["rednote/dots-ocr"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
PaddleOCR-VL vs Nanonets-OCRPaddleOCR-VL vs SuryaGLM-OCR vs dots.ocrQianfan OCR vs dots.ocrPaddleOCR-VL vs DoclingPaddleOCR-VL vs TesseractGoogle Document AI vs dots.ocrPaddleOCR-VL vs Chunkr
PaddleOCR-VL alternativesdots.ocr alternativesBest OCR models in 2026