README.md

---
title: mlx-vision
emoji: 👁️
colorFrom: blue
colorTo: green
sdk: static
pinned: false
license: mit
tags:
  - mlx
  - ocr
  - apple-silicon
  - speculative-decoding
  - dspark
  - deepseek-ocr
  - glm-ocr
  - vision-language-model
  - region:us
library_name: mlx-vision
pipeline_tag: image-to-text
---

# mlx-vision

**DSpark-style faster OCR on Apple Silicon (MLX).**

Speculative decoding for OCR VLMs — DeepSeek-OCR-2, GLM-OCR, Unlimited-OCR.
Vision/OCR counterpart to [mlx-dspark](https://github.com/ARahim3/mlx-dspark).

| Resource | Link |
|----------|------|
| **Code** | [github.com/will702/mlx-vision](https://github.com/will702/mlx-vision) |
| **Agents** | [AGENTS.md](https://github.com/will702/mlx-vision/blob/main/AGENTS.md) · [llms.txt](https://github.com/will702/mlx-vision/blob/main/llms.txt) |
| **Weights used** | [`mlx-community/DeepSeek-OCR-2-8bit`](https://huggingface.co/mlx-community/DeepSeek-OCR-2-8bit), [`GLM-OCR`](https://huggingface.co/mlx-community/GLM-OCR-bf16), [`Unlimited-OCR`](https://huggingface.co/baidu/Unlimited-OCR) |

## Install

```bash
pip install git+https://github.com/will702/mlx-vision.git
# with PDF support:
pip install "mlx-vision[pdf] @ git+https://github.com/will702/mlx-vision.git"
```

## Quick start

```python
from mlx_vision import ocr

result = ocr("document.png", model="deepseek-ocr-2", mode="auto")
print(result.text)
```

```bash
mlx-vision -m deepseek-ocr-2 -i page.png --mode dspark -v
mlx-vision -m deepseek-ocr-2 -i doc.pdf --page 0 --dpi 200
```

## Modes

- `baseline` — mlx-vlm generate
- `dspark` / `lookup` — training-free semi-AR draft + confidence schedule (OCR DSpark weights not public yet)
- `dflash` / `eagle3` / `mtp` — neural speculation via mlx-vlm (`--drafter`)
- `auto` — best available path

## Cite

See [CITATION.cff](https://github.com/will702/mlx-vision/blob/main/CITATION.cff). Related paper: [DSpark](https://arxiv.org/abs/2607.05147).

## License

MIT