docTR vs Typhoon OCR
Pick docTR if you want word-level text with coordinates from a Python library you can train and run yourself, for example on invoices and receipts that feed your own extraction code. Pick Typhoon OCR if you need whole pages converted to Markdown or HTML with tables, especially Thai or Thai-English documents such as government forms and financial reports. The main split is output: docTR returns words and lines only, while Typhoon OCR returns structured documents but covers only Thai and English.
docTRmindee/doctr | Typhoon OCRtyphoon/typhoon-ocr | |
|---|---|---|
| Type | Open source | Open source |
| Price | 1 credit / page | 2 credits / page |
| Per 1,000 pages | $1.00 | $2.00 |
| Typical speed | 0.9 s / page | 1.8 s / page |
| Best for | Word-level OCR with positions | Thai documents |
| Inputs | PDF, images | PDF, images |
| Languages | Mainly Latin scripts | Thai, English |
| Max pages per request | 500 | 500 |
| Tables | No | ✓ Yes |
| Handwriting | No | No |
| Math | No | No |
| Bounding boxes | ✓ Yes | No |
| Confidence | ✓ Yes | No |
| Forms | No | 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 docTR if
- You need word, line and block output with coordinates as JSON, with straight or rotated boxes.
- You want to choose your own detection and recognition architectures and train your own models in Python.
- You want Apache-2.0 with no extra terms, plus ready Docker images and a key information extraction predictor.
Choose Typhoon OCR if
- You need tables as HTML with merged cells, equations as LaTeX and checkboxes marked, in Markdown or HTML output.
- Your documents are Thai or Thai-English: V1.5 reached BLEU 0.870 on Thai government forms, ahead of GPT-5 and Gemini 2.5 Pro.
- You want a small 2B model (0.8B variant also exists) that scores 84.7 on olmOCR-bench, with a hosted API as an option.
Key differences
| docTR | Typhoon OCR | |
|---|---|---|
| What it is | Two-stage text detection and recognition library from Mindee, run in Python on PyTorch | Thai and English document OCR model from Typhoon (SCB 10X), 2B parameters |
| Licence | Apache-2.0 | Apache-2.0, subject to OpenTyphoon Terms and Conditions |
| How you run it | Self-hosted only, on CPU or NVIDIA GPU; GPU Docker images need CUDA 12.2 or newer on the host | Self-hosted weights, or the Typhoon API as typhoon-ocr-v2 |
| Price | Free, you pay for compute | Free self-hosted, you pay for compute. API pricing not published on the model card |
| Output formats | JSON, plain text, bounding boxes. No tables, formulas or Markdown | Markdown, HTML, LaTeX |
| Benchmarks | Not published | V2 2B: olmOCR-bench 84.7, OmniDocBench v1.6 90.97, ThaiOCRBench 59.9 |
| Known failure | Split-crop merging can misjoin repeated characters (issue #2229); cropped images give poorer results (issue #2007); ONNX export of MASTER truncates decoding steps (issue #2224) | V2 can loop until the token limit on dense multi-column pages; key information extraction is weakest at 21.0 on ThaiOCRBench; it can output text not in the image |
Sources
- docTR GitHub README
- docTR documentation
- Issue #2229
- Typhoon OCR V2 2B model card
- Typhoon OCR technical report
What each one is
docTR
Open-source OCR library from Mindee with separate text detection and recognition models.
docTR detailsTyphoon OCR
Open vision model tuned for Thai and English documents, with Markdown output.
Typhoon 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": "mindee/doctr","fallbacks": ["typhoon/typhoon-ocr"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
docTR alternativesTyphoon OCR alternativesBest OCR models in 2026