---
title: OvisOCR2 Document to Markdown
emoji: 📄
colorFrom: indigo
colorTo: green
sdk: gradio
sdk_version: 6.20.0
app_file: app.py
short_description: Parse document images into Markdown with OvisOCR2
python_version: "3.12"
startup_duration_timeout: 30m
---

# OvisOCR2 — Document → Markdown

A Gradio demo for [**ATH-MaaS/OvisOCR2**](https://huggingface.co/ATH-MaaS/OvisOCR2),
a compact 0.9B end-to-end page-level document-parsing / OCR vision-language model
(Qwen3.5 backbone). Upload a document page and the model transcribes it into a single
Markdown document — formatting formulas as LaTeX and tables as HTML.