MiniCPM-OCR-1.0

license: apache-2.0 language:

en
zh pipeline_tag: image-text-to-text tags:
minicpm-ocr
ocr
document-parsing
multimodal
vision-language
table-recognition
formula-recognition base_model:
openbmb/MiniCPM-V-4.6 library_name: minicpm-ocr

Demo (CN / Global) · GitHub · Free API · SDK usage · DSpark · Skill

MiniCPM-OCR 1.0 (1.3B) is the first OCR-specific model in the MiniCPM-V/o family. It focuses on text-centric document understanding — text, tables, formulas, charts, seals, and structured information extraction — and balances quality against speed on real-world captures.

Highlights

🏆 SOTA document parsing: 93.68 on Real5-OmniDocBench, a benchmark for real-world document parsing, with leading results on OmniDocBench v1.6, Wild-OmniDocBench, and olmOCR-Bench. Handles Chinese and English documents and PDFs, including scans, curved pages, photographs, skewed pages, and recaptured documents.
⚡ Efficient inference. With DSpark decoding acceleration (BF16, GGUF) end-to-end processing efficiency improves by 55%. Throughput is about 1.49× PaddleOCR-VL-1.6 and 3.62× MinerU2.5 Pro.
🔍 General recognition and information extraction: 54.7 on OCRBench v2, plus key information extraction (KIE) for downstream automation.
🛠️ Easy to deploy and develop: vLLM, SGLang, llama.cpp, and Ollama from day one, with quantized variants and multi-chip builds. Supports further fine-tuning, and ships a pip SDK, a full API, and agent workflows. Online demos: HF Spaces · CN · Official CN · Official Global.

SOTA quality at higher throughput

Among the open-source OCR models compared, MiniCPM-OCR 1.0 reaches 1.2 pages/s and the highest equal-weight average score across OmniDocBench v1.6, Real5-OmniDocBench, Wild-OmniDocBench, and olmOCR-Bench.

Average score is the equal-weight mean of the four benchmarks. All models use accelerated configurations. Settings and reproduction: reproduction guide. Per-model throughput: model guide.

Benchmark comparisons

Full MiniCPM-OCR 1.0 evaluation results

Each panel keeps its own benchmark's scores; higher is better. Scores from different benchmarks are not directly comparable. Test settings and reproduction steps are in the reproduction guide; per-benchmark discussion is in the model guide.

Throughput

On an NVIDIA GeForce RTX 5090 with all 1651 OmniDocBench v1.6 images, image-level concurrency 1, and each system's accelerated configuration, MiniCPM-OCR 1.0 reaches 1.2 pages/s (relative throughput 1.00×; higher is faster).

Concurrency 1 is image-level: regions inside a page are still recognised in parallel. Real throughput also depends on page content, backend, concurrency, and hardware.

Examples

Left to right: original document, detected layout, and rendered Markdown.

English documents

Chinese documents

Original documents and parsed results, side by side.

English documents

Chinese documents

Model architecture

Item	Description
Parameters	1.3B
Base	Trained for OCR on top of MiniCPM-V 4.6
Components	SigLIP vision encoder + Qwen3.5 language model
Primary languages	Chinese and English
Tasks	Text, tables, formulas, charts, seals, and key information extraction
Parsing pipeline	MiniCPM-OCR 1.0 + PP-DocLayoutV3 layout model + `minicpmvocr` SDK
Visual tokens	4× compression, `scale_resolution` 448, up to 9 slices

See the model guide for architecture, parsing workflow, deployment options, and benchmark breakdowns.

Quick start

Online Demo

To quickly try text-centric document understanding and structured information extraction, we recommend HF Spaces, CN, Official CN, Official Global / API.

Use with the MiniCPM-OCR SDK

Required before using either the CLI or Python API: deploy and start the OCR model service first, and wait until it is ready to accept requests. The SDK examples below are clients of that running service; they do not deploy or start the OCR model service automatically.

Use the official minicpm-ocr SDK to parse a local image or PDF. It detects layout with PP-DocLayoutV3, recognises each region, and saves structured JSON and Markdown in reading order.

Install Python 3.10+ and the SDK:

pip install "minicpm-ocr[all]"

Install minicpm-ocr for your platform using the installation guide.

Prerequisite: deploy and start the model service

Before running either the CLI or Python API, deploy and start a vLLM recognition service for openbmb/MiniCPM-OCR-1.0 following the deployment guide. The examples assume a service at http://127.0.0.1:8080, served model name minicpm-ocr, and an NVIDIA GPU for layout detection. Match the endpoint, backend, and device to your deployment. Configure the service to listen on port 8080 and serve the model as minicpm-ocr to use the examples unchanged.

CLI

Start the model service described above and wait until it is ready before running this command. Save a local image as demo.png, then run:

minicpmvocr parse demo.png \
  --backend vllm \
  --server-url http://127.0.0.1:8080 \
  --model minicpm-ocr \
  --device cuda:0 \
  --set run.run_root=output \
  --output output/demo.md

The CLI automatically downloads PP-DocLayoutV3 to models/PP-DocLayoutV3_infer/ when needed. The final Markdown is saved to output/demo.md; structured JSON and per-page Markdown are kept in output/<run_tag>/results_json/ and output/<run_tag>/output/.

To parse a PDF, replace demo.png with its local path. Use --device cpu for CPU layout detection. Run minicpmvocr parse --help for all CLI options.

Python API

The model service described above must already be running before you execute this Python example. Set engine.server_url and engine.model_name to match that service.

Save a local image as demo.png, then run:

from importlib.resources import files
from pathlib import Path

from minicpmvocr import load_config, pipelined_run
from minicpmvocr.layouts.download import ensure_layout_model


def main():
    layout_model = ensure_layout_model(Path("models/PP-DocLayoutV3_infer"))

    cfg = load_config(
        str(files("minicpmvocr").joinpath("configs/minicpmvocr.yaml")),
        overrides={
            "input": {"file_path": str(Path("demo.png").resolve())},
            "engine": {
                "type": "vllm",
                "server_url": "http://127.0.0.1:8080",
                "model_name": "minicpm-ocr",
            },
            "layout": {
                "model_dir": str(layout_model),
                "device": "cuda:0",
            },
            "pipeline": {"overlap_layout": True},
            "run": {"run_root": str(Path("output").resolve())},
        },
    )

    pipelined_run(cfg)

    print("JSON directory:", cfg["_meta"]["result_json_dir"])
    print("Markdown directory:", cfg["_meta"]["output_dir"])


if __name__ == "__main__":
    main()

ensure_layout_model() downloads the official layout model when it is missing. Results are written to output/<run_tag>/results_json/ and output/<run_tag>/output/. To parse a PDF, replace demo.png with its local path.

For a CPU layout detector, set layout.device to cpu. For the GGUF checkpoint served by llama.cpp, set engine.type to llamacpp and match the service context length with engine.max_model_len.

See the SDK API reference for configuration and document-level Markdown export.

Agent Skill

You can also have a coding agent follow the Agent Skill to set up the environment, download the model, and parse a first document.

Read and execute https://github.com/OpenBMB/MiniCPM-OCR/blob/main/docs/SKILL_en.md

Fine-tuning

LLaMA-Factory · ms-swift — data preparation, full-parameter SFT with the multimodal projector frozen, training, and evaluation (French handwritten text lines).

Recognition guide

Related checkpoints

Format	Hugging Face	ModelScope
BF16 (original) (this repo)	🤗	ModelScope
GGUF (llama.cpp)	🤗	ModelScope
AWQ W4A16	🤗	ModelScope
GPTQ W4A16	🤗	ModelScope
DSpark draft (BF16)	🤗	ModelScope
DSpark draft (GGUF)	🤗	ModelScope

DSpark requires a matching main and draft model from the same release — do not mix draft weights with a different main checkpoint. See DSpark acceleration.

For OCR quality assessment (auto-accept, human-review routing, best-of-N across models), see MiniCPM-OCR Jev.

FAQ

Installation, model downloads, GPU memory, input formats, and unexpected outputs: FAQ.

License

Model weights, code, and documentation are released under the Apache License 2.0.