Unlimited OCR vs docTR
Pick Unlimited OCR to turn long PDFs, reports and books into Markdown with bounding boxes in one request, if you have an NVIDIA GPU. Pick docTR for word-level OCR with coordinates on invoices and receipts, or when you need to run on CPU and train your own models. The main split is output: Unlimited OCR parses whole documents to Markdown, while docTR returns only words and lines.
Unlimited OCRbaidu/unlimited-ocr | docTRmindee/doctr | |
|---|---|---|
| Type | Open source | Open source |
| Price | 2 credits / page | 1 credit / page |
| Per 1,000 pages | $2.00 | $1.00 |
| Typical speed | 1.2 s / page | 0.9 s / page |
| Best for | Books and very long documents | Word-level OCR with positions |
| Inputs | PDF, images | PDF, images |
| Languages | 30+ languages | Mainly Latin scripts |
| Max pages per request | 1,000 | 500 |
| Tables | ✓ Yes | No |
| Handwriting | No | No |
| Math | ✓ Yes | No |
| Bounding boxes | ✓ Yes | ✓ Yes |
| Confidence | No | ✓ Yes |
| Forms | No | 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 Unlimited OCR if
- It parses 40+ pages in a single forward pass with roughly constant memory, so a long report or book goes in as one request.
- It outputs Markdown, plain text and bounding boxes with block types, which suits bulk PDF to text for RAG.
- It is MIT licensed, and Baidu Cloud also offers a hosted endpoint if you do not want to run a GPU.
Choose docTR if
- It runs on CPU or NVIDIA GPU, while Unlimited OCR needs an NVIDIA GPU with CUDA 12.9 or 13.0.
- It exports a nested page, block, line and word structure with coordinates as JSON, and lets you choose detection and recognition architectures with straight or rotated boxes.
- It is Apache-2.0, ships Docker images, and includes a key information extraction predictor for invoices and receipts feeding custom extraction.
Key differences
| Unlimited OCR | docTR | |
|---|---|---|
| What it is | 3B Mixture-of-Experts document parser from Baidu, derived from DeepSeek-OCR | Two-stage text detection and recognition library from Mindee, run in Python on PyTorch |
| Licence | MIT | Apache-2.0 |
| How you run it | Self-host with vLLM, SGLang or transformers, or use the hosted endpoint in Baidu Cloud's OCR product | Local Python library with Docker images; GPU images need CUDA 12.2 or newer on the host |
| Hardware | NVIDIA GPU with CUDA 12.9 or 13.0; Unlimited OCR 2 on SGLang runs out of memory on 12 GB GPUs; not for CPU-only machines | CPU or NVIDIA GPU |
| Output formats | Markdown, plain text, bounding boxes | JSON, plain text, bounding boxes; no table, formula or Markdown output |
| Benchmarks | ParseBench mean 46.17 (independent). Unlimited OCR 2: MPDocBench 84.6 and MarathonDocBench 77.1 (authors) | Not published |
| Known failure | Text repetition and looping, hallucinated text, and a 32,768 token context cap; multi-page input only works in base mode | Split-crop merging misjoins repeated characters, cropping images gives poorer results, and ONNX export of MASTER truncates decoding steps |
Sources
- Hugging Face model card
- GitHub repo
- MarkTechPost on Unlimited OCR 2
- docTR GitHub README
- docTR documentation
What each one is
Unlimited OCR
Open-source model from Baidu that reads hundreds of pages in one pass with flat memory use.
Unlimited OCR detailsdocTR
Open-source OCR library from Mindee with separate text detection and recognition models.
docTR 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": "baidu/unlimited-ocr","fallbacks": ["mindee/doctr"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
Unlimited OCR alternativesdocTR alternativesBest OCR models in 2026