---
title: "Tencent's WeVisDoc-4B Tops Document Parsing Benchmarks Without OCR or Detectors"
subtitle: "Tencent's WeChat Vision team released a compact 4B document parser that tops OmniDocBench and PureDocBench among end-to-end systems, with Apache-2.0 weights."
url: "https://alphasignal.ai/news/tencent-s-wevisdoc-4b-tops-document-parsing-benchmarks-without-ocr-or-detectors"
author: "AlphaSignal Newsroom"
author_url: "https://alphasignal.ai/editorial-team"
company: "Tencent Hy"
company_url: "https://alphasignal.ai/company/tencenthunyuan"
published: "2026-09-16T03:07:01.000Z"
updated: "2026-09-24T13:04:33.492Z"
type: "Model"
topics:
  - "Image"
  - "Open Source"
subtopics:
  - "Ocr"
upvotes: 115
source_url: "https://huggingface.co/tencent/WeVisDoc-4B"
comments: 0
truncated: true
---

# Tencent's WeVisDoc-4B Tops Document Parsing Benchmarks Without OCR or Detectors

> Tencent's WeChat Vision team released a compact 4B document parser that tops OmniDocBench and PureDocBench among end-to-end systems, with Apache-2.0 weights.

Tencent’s WeChat Vision team has released [WeVisDoc-4B](https://huggingface.co/tencent/WeVisDoc-4B), an open-weights document parser that converts page images into structured Markdown, with LaTeX for equations and HTML for tables. The authors’ evaluations place the 4-billion-parameter model first among the listed end-to-end parsers on OmniDocBench v1.6 and all three PureDocBench tracks.

The accompanying [project repository](https://github.com/Tencent/WeVisDoc) describes a two-stage training framework built on unmodified Qwen3-VL-Instruct 2B and 4B backbones. Inference requires no separate detector, OCR engine, table parser, or custom decoder. Both checkpoints use the Apache 2.0 license, which permits commercial deployment subject to its terms.

## Top scores, including damaged pages

On the authors’ reported evaluations, where higher scores indicate better parsing, WeVisDoc-4B reaches 95.38 on OmniDocBench v1.6 and averages 75.54 across PureDocBench’s three tracks. The 2B checkpoint follows closely despite its smaller parameter count.

__Reported overall benchmark scores__
| Model       | OmniDocBench v1.6 | PureDocBench mean |
| ----------- | ----------------- | ----------------- |
| WeVisDoc-2B | 95.06             | 73.86             |
| WeVisDoc-4B | 95.38             | 75.54             |

![Benchmark comparison of WeVisDoc-4B and other end-to-end document parsers](https://imagedelivery.net/kokADdxuw1FCXyR2p8D2CQ/f878c48b-611f-4a45-2ee5-282bbf805a00/public) 

The project’s comparison covers OmniDocBench v1.6 and the three PureDocBench tracks.

PureDocBench separates clean digital pages from two harder categories: Digital Degraded includes compression artifacts and low-resolution inputs, while Real Degraded includes photographed pages, skew, and shadows. WeVisDoc-4B scores 79.81 on Clean, 77.74 on Digital Degraded, and 69.08 on Real Degraded.

_Preview only. Full text requires an AlphaSignal Pro account._

## Sources

- [tencent/WeVisDoc-4B on Hugging Face](https://huggingface.co/tencent/WeVisDoc-4B)
- [WeVisDoc: From Coverage to Capability for Robust End-to-End Document Parsing](https://arxiv.org/html/2609.20423v1)
- [Tencent/WeVisDoc GitHub repository](https://github.com/Tencent/WeVisDoc)
- [DailyPapers announcement tweet](https://twitter.com/HuggingPapers/status/2100609544946643118)