docTR vs MiniCPM-V 4.5
Pick docTR when you need word-level text with coordinates from invoices and receipts and plan to feed your own extraction or train your own models in Python. Pick MiniCPM-V 4.5 when you want a local model that returns Markdown or JSON and can answer questions about a document on a single GPU or edge device. The main split is output: docTR gives words and boxes, MiniCPM-V gives text without layout or coordinates.
docTRmindee/doctr | MiniCPM-V 4.5openbmb/minicpm-v | |
|---|---|---|
| Type | Open source | Vision model |
| Price | 1 credit / page | 2 credits / page |
| Per 1,000 pages | $1.00 | $2.00 |
| Typical speed | 0.9 s / page | 2.5 s / page |
| Best for | Word-level OCR with positions | OCR on edge devices |
| Inputs | PDF, images | images |
| Languages | Mainly Latin scripts | 30+ languages |
| Max pages per request | 500 | 500 |
| Tables | No | ✓ Yes |
| Handwriting | No | ✓ Yes |
| 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 nested page, block, line and word JSON with coordinates for each word.
- You want to pick your own detection and recognition architectures, with straight or rotated boxes.
- You train your own models in Python, or want a key information extraction predictor under Apache-2.0.
Choose MiniCPM-V 4.5 if
- You want Markdown or JSON output from a document instead of words and lines.
- You need quantized on-device or edge OCR, with int4, GGUF and AWQ builds on llama.cpp and vLLM.
- You want local document Q&A on a single GPU.
Key differences
| docTR | MiniCPM-V 4.5 | |
|---|---|---|
| Type | Two-stage text detection and recognition library from Mindee, run in Python on PyTorch | 8.7B open-weight vision language model from OpenBMB, built on Qwen3-8B and SigLIP2-400M |
| Licence and price | Apache-2.0, free self-hosted, you pay for compute | Apache-2.0 weights and code, free self-hosted, you pay for compute |
| How you run it | Python library, plus Docker images; GPU images need CUDA 12.2 or newer on the host | llama.cpp, vLLM, Ollama or Transformers; Ollama and SGLang support were still pending official releases on the model card |
| Output formats | JSON, plain text, bounding boxes | Plain text, Markdown, JSON |
| Coordinates and layout | Word, line and block coordinates, straight or rotated boxes | No bounding box or layout output |
| Tables and Markdown | No table, formula or Markdown output; words and lines only | Markdown output; claims state of the art PDF parsing among general MLLMs on OmniDocBench |
| Benchmarks | Not published | OpenBMB claims leading OCRBench results, but the model card gives scores only as images, so numbers are hard to verify |
Sources
What each one is
docTR
Open-source OCR library from Mindee with separate text detection and recognition models.
docTR detailsMiniCPM-V 4.5
8B vision model from OpenBMB (August 2025) built to run on devices, with strong OCR and document parsing.
MiniCPM-V 4.5 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": ["openbmb/minicpm-v"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
docTR alternativesMiniCPM-V 4.5 alternativesBest OCR models in 2026