Typhoon OCR vs MiniCPM-V 4.5
Pick Typhoon OCR for Thai and English documents such as government forms, financial reports and books, where it has published scores and tables come out as HTML. Pick MiniCPM-V 4.5 if you want a general 8.7B vision model that you run locally for on-device OCR or document Q&A. The main split is scope: Typhoon OCR is a 2B document specialist for two languages, MiniCPM-V is a general model with no published benchmark numbers in text.
Typhoon OCRtyphoon/typhoon-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 | 1.8 s / page | 2.5 s / page |
| Best for | Thai documents | OCR on edge devices |
| Inputs | PDF, images | images |
| Languages | Thai, English | 30+ languages |
| Max pages per request | 500 | 500 |
| Tables | ✓ Yes | ✓ Yes |
| Handwriting | No | ✓ Yes |
| Math | No | No |
| Bounding boxes | No | 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 Typhoon OCR if
- It scores 84.7 on olmOCR-bench and 90.97 on OmniDocBench v1.6 at 2B parameters, with the figures published as numbers.
- It is built for Thai government forms, financial reports, Thai books and Thai-English mixed pages; V1.5 scored BLEU 0.870 on Thai government forms, ahead of GPT-5 and Gemini 2.5 Pro.
- It outputs tables as HTML with merged cells, equations as LaTeX and marks checkboxes, and you can also call it through the Typhoon API as typhoon-ocr-v2.
Choose MiniCPM-V 4.5 if
- It is fully self-hosted under Apache-2.0 and runs on llama.cpp and vLLM, with int4, GGUF and AWQ builds for edge and on-device use.
- It is a general vision language model, so it can answer questions about a document locally on a single GPU, not only transcribe it.
- It is not limited to Thai and English the way Typhoon OCR's evaluation is, and it returns plain text, Markdown or JSON.
Key differences
| Typhoon OCR | MiniCPM-V 4.5 | |
|---|---|---|
| Model size | 2B (V2 and V1.5), plus a 0.8B V2 variant | 8.7B, built on Qwen3-8B and SigLIP2-400M |
| Licence | Apache-2.0, subject to OpenTyphoon Terms and Conditions | Apache-2.0 weights and code |
| How you run it | Self-host the weights or use the Typhoon API (typhoon-ocr-v2) | Self-host only, with llama.cpp, vLLM, Ollama or Transformers |
| Output formats | Markdown, HTML, LaTeX | Plain text, Markdown, JSON |
| Published benchmarks | olmOCR-bench 84.7, OmniDocBench v1.6 90.97, ThaiOCRBench 59.9 (V2 2B) | Scores appear only as images on the model card, so numbers are hard to verify |
| Known failure | V2 can loop until the token limit on dense multi-column pages; key-information extraction scores 21.0 on ThaiOCRBench | No bounding box or layout output; Ollama and SGLang support were still pending on the model card |
| Hosted option and price | Free to self-host; Typhoon API pricing not listed | Free to self-host; no hosted service |
Sources
- Typhoon OCR V2 2B model card
- Typhoon OCR technical report
- Typhoon OCR 1.5 model card
- MiniCPM-V 4.5 Hugging Face model card
- MiniCPM-V GitHub repo
What each one is
Typhoon OCR
Open vision model tuned for Thai and English documents, with Markdown output.
Typhoon 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": "typhoon/typhoon-ocr","fallbacks": ["openbmb/minicpm-v"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
Typhoon OCR alternativesMiniCPM-V 4.5 alternativesBest OCR models in 2026