Pick

Tencent Releases Document Parser WeVisDoc

·2026.09.18 00:23

Key point

Tencent has released WeVisDoc, a document parser based on Qwen3-VL, which ranked first on OmniDocBench.

Details

Tencent has released WeVisDoc, an end-to-end document parser that converts page images into structured Markdown. The model was developed by fine-tuning Qwen3-VL-2B-Instruct and Qwen3-VL-4B-Instruct, generating accurate document structures including LaTeX formulas and HTML tables.

WeVisDoc-4B achieved an overall score of 95.38 on OmniDocBench v1.6 and an average score of 75.54 across three tracks on PureDocBench. This performance ranks 1st in all four settings among the compared end-to-end parsers.

computer-vision open-source tencent document-parsing qwen3-vl omnidocbench

Read originalShare on Threads

This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.