Nougat vs MiniCPM-V 4.5
Pick Nougat for English arXiv-style papers where you need LaTeX for equations and tables, and only for non-commercial work. Pick MiniCPM-V 4.5 for anything commercial, for a general OCR and document Q&A model on one GPU, or for quantized edge use. The main reason is the licence: Nougat's weights are CC-BY-NC, while MiniCPM-V 4.5 is Apache-2.0.
Nougatmeta/nougat | MiniCPM-V 4.5openbmb/minicpm-v | |
|---|---|---|
| Type | Open source | Vision model |
| Price | 2 credits / page | 2 credits / page |
| Per 1,000 pages | $2.00 | $2.00 |
| Typical speed | 3.0 s / page | 2.5 s / page |
| Best for | arXiv-style papers | OCR on edge devices |
| Inputs | PDF, images | images |
| Languages | Mainly English | 30+ languages |
| Max pages per request | 200 | 500 |
| Tables | ✓ Yes | ✓ Yes |
| Handwriting | No | ✓ Yes |
| Math | ✓ Yes | No |
| Bounding boxes | No | No |
| Confidence | No | 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 Nougat if
- You are converting English scientific papers, since it was trained on arXiv and PubMed Central documents.
- You need equations and tables as LaTeX inside Mathpix-compatible .mmd Markdown.
- You want a small or base checkpoint with a simple CLI and API.
Choose MiniCPM-V 4.5 if
- You are building a commercial product, because the weights and code are Apache-2.0.
- You need to run on-device or on a single GPU, using llama.cpp or vLLM with int4, GGUF or AWQ builds.
- You want local document Q&A with plain text, Markdown or JSON output, not just a page-to-Markdown converter.
Key differences
| Nougat | MiniCPM-V 4.5 | |
|---|---|---|
| Licence | Code MIT; weights CC-BY-NC, so no commercial use | Apache-2.0 for weights and code |
| What it is | Meta AI Visual Transformer for scientific PDF pages | 8.7B open-weight vision language model from OpenBMB, built on Qwen3-8B and SigLIP2-400M |
| How you run it | Self-hosted Python package with a CLI and API; small and base checkpoints | Self-hosted with llama.cpp, vLLM or Transformers; int4, GGUF and AWQ builds. Ollama and SGLang support were still pending official releases |
| Output formats | Markdown (.mmd) and LaTeX | Plain text, Markdown and JSON |
| Languages | Works best on English; Chinese, Russian and Japanese do not work | Not published |
| Known failure | Repetition loops on 1.5% of test pages in the paper, more on out-of-domain documents; repetition detection can drop pages as MISSING_PAGE | Benchmark scores are given only as images, so they are hard to verify. No int4 VRAM figure published |
| Layout and coordinates | Not published | No bounding box or layout output |
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 detailsTry both on your own file
Run the file once with each model and compare the output.
PDF, PNG, JPG, WebP, HEIC. Up to 10 files at onceNo file?
Use one, fall back to the other
{"model": "meta/nougat","fallbacks": ["openbmb/minicpm-v"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
Nougat vs DeepSeek-VL2Nougat vs InternVL3.5Donut vs MiniCPM-V 4.5MonkeyOCR vs MiniCPM-V 4.5Google Document AI vs NougatKraken vs MiniCPM-V 4.5Nougat vs Phi-4-multimodalMinerU vs MiniCPM-V 4.5
Nougat alternativesMiniCPM-V 4.5 alternativesBest OCR models in 2026