OCRRouter

Mistral OCR vs Hunyuan OCR

Pick Mistral OCR for high-volume PDF to Markdown and scanned business documents, where a hosted API at $2 per 1,000 pages ($1 via Batch) means no GPU to run. Pick Hunyuan OCR if you must self-host, or need photo and scene text spotting, image translation or historical scripts, and you are outside the EU, UK and South Korea. Check Mistral output on dense tables and handwriting, and note that Hunyuan's licence excludes several regions.

Mistral OCRmistral/mistral-ocrHunyuan OCRtencent/hunyuan-ocr
TypeOCR APIOpen source
Price25 credits / page2 credits / page
Per 1,000 pages$25.00$2.00
Typical speed0.9 s / page1.5 s / page
Best forHigh-volume PDF to MarkdownChinese documents and photos
InputsPDF, imagesPDF, images
Languages100+ languages100+ languages
Max pages per request1,000500
Tables✓ Yes✓ Yes
HandwritingNoNo
Math✓ Yes✓ Yes
Bounding boxes✓ Yes✓ Yes
ConfidenceNoNo
FormsNoNo
Multilingual✓ Yes✓ Yes
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 Mistral OCR if

  • You want a hosted API with no GPU to manage and a list price of $2 per 1,000 pages.
  • You process large batches of scans, since the Batch API halves OCR to $1 per 1,000 pages.
  • You need tables rebuilt as HTML with rowspan and colspan, with Markdown, HTML, JSON and extracted images as output.

Choose Hunyuan OCR if

  • You need to run the model yourself. It is about 1B parameters and has a llama.cpp GGUF path for CPU, consumer GPU or laptop.
  • Your work is photo and scene text spotting, text-image translation, or low-resource and ancient scripts.
  • You want one model for parsing, spotting, extraction and translation, with bounding boxes, LaTeX and JSON output and images up to 4K.

Key differences

Mistral OCRHunyuan OCR
How you run itHosted API (current pinned version mistral-ocr-2512, OCR 3)Self-hosted via vLLM, Transformers or llama.cpp
LicenceProprietaryTencent Hunyuan Community License. It excludes the EU, UK and South Korea, and above 100M monthly active users you need a separate licence
Price$2 per 1,000 pages; $3 with structured annotations; $1 via Batch APIFree software; you pay for compute
Output formatsMarkdown, HTML, JSON, imagesMarkdown, HTML, LaTeX, JSON, plain text, bounding boxes
HardwareNone on your sideNVIDIA GPU with CUDA 13 recommended for the unified vLLM environment (vLLM 0.25.1+); GGUF path for CPU or laptop
Independent benchmarkOmniDocBench v1.6 overall: 85.66 (v1.5 score of 78.83 was for the first version, 2503)ParseBench-2 overall (v1.5): 67.11; ParseBench mean: 63.86
Known failure modesHallucinated text on unreadable handwriting; colored-background pages returned as images with no text (2025 release); changed checkbox states and misread characters on a dense financial tableArchived 1.0 note says Transformers and vLLM outputs differed in accuracy; text score of 62.88 on ParseBench-2 in public CI replay

Sources

What each one is

Mistral OCR

Fast document OCR that returns clean Markdown with tables and embedded images.

Mistral OCR details

Hunyuan OCR

Small end-to-end OCR model from Tencent covering parsing, extraction and photo translation.

Hunyuan OCR 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": "mistral/mistral-ocr","fallbacks": ["tencent/hunyuan-ocr"],"file_url": "https://example.com/document.pdf"}

Questions

More comparisons

Mistral OCR alternativesHunyuan OCR alternativesBest OCR models in 2026