MiniCPM-V 4.5 vs Chunkr
Pick MiniCPM-V 4.5 if you want a free Apache-2.0 model you run yourself for plain text, Markdown or JSON output, for example local document Q&A on one GPU or quantized on-device OCR. Pick Chunkr if you need layout analysis, bounding boxes and chunked output from PDFs, slides and Word files, for example to feed a RAG pipeline. The main split is that MiniCPM-V is a general vision language model with no coordinates, while Chunkr is a document parsing service with layout and boxes.
MiniCPM-V 4.5openbmb/minicpm-v | Chunkrchunkr/chunkr | |
|---|---|---|
| Type | Vision model | OCR API |
| Price | 2 credits / page | 14 credits / page |
| Per 1,000 pages | $2.00 | $14.00 |
| Typical speed | 2.5 s / page | 3.0 s / page |
| Best for | OCR on edge devices | Layout-aware chunks for RAG |
| Inputs | images | PDF, images, DOCX |
| Languages | 30+ languages | Multilingual |
| Max pages per request | 500 | 1,000 |
| Tables | ✓ Yes | ✓ Yes |
| Handwriting | ✓ Yes | No |
| Math | No | ✓ Yes |
| Bounding boxes | No | ✓ Yes |
| Confidence | No | 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 MiniCPM-V 4.5 if
- You want Apache-2.0 weights and code with no commercial license question for closed products.
- You need to run quantized OCR on a device or a single GPU, using int4, GGUF or AWQ builds on llama.cpp or vLLM.
- You want one model that reads a document and answers questions about it, and you do not need an external LLM endpoint.
Choose Chunkr if
- You need OCR with bounding boxes, which MiniCPM-V does not output.
- You parse PDFs, PPT, Word files and images into HTML, Markdown or JSON, and want semantic chunking for RAG.
- You want a choice between a hosted cloud API and a self-hosted Docker setup with GPU, CPU-only or Mac ARM options.
Key differences
| MiniCPM-V 4.5 | Chunkr | |
|---|---|---|
| Licence | Apache-2.0 for weights and code | AGPL-3.0 or commercial license; closed-source use needs the commercial license |
| How you run it | Self-host only, with llama.cpp, vLLM, Ollama or Transformers. Ollama and SGLang support were still pending official releases on the model card. | Hosted cloud API, or self-host via Docker. Self-host needs an external OpenAI-compatible LLM endpoint. |
| Price | Free, you pay for compute. No int4 VRAM figure published. | Hosted: credit-based plans, official prices not verified. Open-source version is free, you pay for compute and an LLM endpoint. |
| Output formats | Plain text, Markdown, JSON | HTML, Markdown, JSON, bounding boxes |
| Layout and coordinates | General VLM with no bounding box or layout output | Layout analysis, OCR with bounding boxes and semantic chunking |
| Input files | Not published | PDF, PPT, Word and images. Excel is not supported in the open-source version. |
| Accuracy evidence | OpenBMB claims leading OCRBench results and state of the art PDF parsing on OmniDocBench among general MLLMs. Scores are given only as images on the model card, so they are hard to verify. | Not published. The open-source version uses community models that are less accurate than the cloud's in-house models. |
Sources
What each one is
MiniCPM-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 detailsChunkr
Document parsing API that segments and chunks pages for RAG. Self-hostable version available.
Chunkr 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": "openbmb/minicpm-v","fallbacks": ["chunkr/chunkr"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
MiniCPM-V 4.5 alternativesChunkr alternativesBest OCR models in 2026