Hunyuan OCR vs MiniCPM-V 4.5
Pick Hunyuan OCR for photo and scene text spotting, text-image translation, and low-resource or historical scripts, and when you need bounding boxes. Pick MiniCPM-V 4.5 for local document Q&A and on-device OCR under Apache-2.0, or when the Hunyuan licence rules you out. The main split is licence and output: Hunyuan is a ~1B model with a regional licence and box output, MiniCPM-V is an 8.7B general VLM with an open licence and no coordinates.
Hunyuan OCRtencent/hunyuan-ocr | 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 | 1.5 s / page | 2.5 s / page |
| Best for | Chinese documents and photos | OCR on edge devices |
| Inputs | PDF, images | images |
| Languages | 100+ languages | 30+ languages |
| Max pages per request | 500 | 500 |
| Tables | ✓ Yes | ✓ Yes |
| Handwriting | No | ✓ Yes |
| Math | ✓ Yes | No |
| Bounding boxes | ✓ Yes | No |
| 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 Hunyuan OCR if
- You need bounding boxes, since Hunyuan OCR outputs them alongside Markdown, HTML, LaTeX, JSON and plain text.
- You work with photo and scene text, text-image translation, or low-resource and ancient scripts, which it covers through 12 task types in the vLLM client.
- You want a small model: about 1B parameters, with a llama.cpp GGUF path that runs on a laptop, consumer GPU or CPU.
Choose MiniCPM-V 4.5 if
- You need a licence with no regional limits or user caps, because MiniCPM-V 4.5 weights and code are Apache-2.0.
- You want to ask questions about documents on a single GPU, or run quantized OCR on-device with int4, GGUF or AWQ builds.
- You prefer a general VLM that runs on llama.cpp, vLLM, Ollama or Transformers, and OpenBMB claims leading OCRBench and OmniDocBench results for it.
Key differences
| Hunyuan OCR | MiniCPM-V 4.5 | |
|---|---|---|
| Licence | Tencent Hunyuan Community License; excludes the EU, UK and South Korea; separate licence above 100M monthly active users | Apache-2.0 |
| Size | About 1B parameters | 8.7B, built on Qwen3-8B and SigLIP2-400M |
| Output formats | Markdown, HTML, LaTeX, JSON, plain text, bounding boxes | Plain text, Markdown, JSON; no bounding box or layout output |
| Runtime | vLLM, Transformers or llama.cpp; unified vLLM environment needs CUDA 13 and vLLM 0.25.1+ | llama.cpp, vLLM, Ollama or Transformers; Ollama and SGLang support were still pending on the model card |
| Published benchmarks | ParseBench-2 overall 67.11, ParseBench mean 63.86 (v1.5) | Scores are given only as images on the model card, so numbers are hard to verify |
| Input and context | Images up to 4K and a 128K context window in v1.5 | Not published |
| Known issues | Archived 1.0 note said Transformers and vLLM outputs differed in accuracy; text score 62.88 on ParseBench-2 | No int4 VRAM figure published |
Sources
What each one is
Hunyuan OCR
Small end-to-end OCR model from Tencent covering parsing, extraction and photo translation.
Hunyuan OCR 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": "tencent/hunyuan-ocr","fallbacks": ["openbmb/minicpm-v"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
Hunyuan OCR alternativesMiniCPM-V 4.5 alternativesBest OCR models in 2026