Florence-2 vs MiniCPM-V 4.5
Pick MiniCPM-V 4.5 for full-page documents, Markdown or JSON output, and any CJK text. Pick Florence-2 when you need a tiny model that returns per-line boxes for scene text and short English labels. The split is output: Florence-2 gives text and boxes only, MiniCPM-V gives structured text but no coordinates.
Florence-2microsoft/florence-2 | 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 | 0.8 s / page | 2.5 s / page |
| Best for | Lightweight OCR with boxes | OCR on edge devices |
| Inputs | images | images |
| Languages | English | 30+ languages |
| Max pages per request | 500 | 500 |
| Tables | No | ✓ Yes |
| Handwriting | No | ✓ Yes |
| Math | No | No |
| Bounding boxes | ✓ Yes | 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 Florence-2 if
- It is 0.23B (base) or 0.77B (large), far smaller than MiniCPM-V's 8.7B, so it suits edge hardware.
- OCR_WITH_REGION returns quadrilateral boxes per text line, which MiniCPM-V does not provide.
- One model also does captioning, detection and grounding, so a single pipeline can cover more than OCR.
Choose MiniCPM-V 4.5 if
- It outputs Markdown and JSON, while Florence-2 gives only plain text and boxes with no tables or LaTeX.
- OpenBMB claims state of the art PDF parsing among general MLLMs on OmniDocBench and leading OCRBench results.
- It runs on llama.cpp and vLLM with int4, GGUF and AWQ builds, which fits local document Q&A on a single GPU.
Key differences
| Florence-2 | MiniCPM-V 4.5 | |
|---|---|---|
| Size | 0.23B (base), 0.77B (large) | 8.7B |
| Licence | MIT | Apache-2.0 for weights and code |
| Output formats | Plain text and bounding boxes | Plain text, Markdown and JSON |
| Coordinates | Quadrilateral boxes per text line via OCR_WITH_REGION | No bounding box or layout output |
| Document structure | No Markdown, tables or LaTeX | Claimed state of the art PDF parsing on OmniDocBench |
| Chinese and handwriting | Chinese output mostly wrong because the vocabulary lacks CJK characters. Spelling errors and missing spaces on cursive handwriting and French text in one blog test | Not published |
| Benchmarks | TextCaps CIDEr 72.8 zero-shot (large), by the authors | Scores given only as images on the model card, so hard to verify |
Sources
- Florence-2-large model card
- HF discussion Chinese very bad
- HF blog OCR test
- Hugging Face model card
- MiniCPM-V 4.5 paper
What each one is
Florence-2
Small Microsoft vision model (0.23B and 0.77B) with OCR and OCR-with-regions tasks that return text with boxes.
Florence-2 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": "microsoft/florence-2","fallbacks": ["openbmb/minicpm-v"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
Florence-2 alternativesMiniCPM-V 4.5 alternativesBest OCR models in 2026