Tencent open-sourced WeVisDoc-2B, a document parser that converts page images to Markdown
Tencent released WeVisDoc-2B, an end-to-end document parser fine-tuned from Qwen3-VL-2B-Instruct, on Hugging Face under Apache-2.0. It achieves an Overall score of 95.06 on OmniDocBench v1.6 and a mean Overall score of 73.86 across PureDocBench tracks.
China context
- Original name
- 腾讯
- Outside China
- Open weights · huggingface.co
- Claims
- Company-reported; not yet independently evaluated
- For builders
- Builders outside China can download the model from Hugging Face and integrate it into document processing pipelines under Apache-2.0.
- For investors
- Investors can monitor adoption of WeVisDoc-2B in open-source document processing projects as a signal of Tencent's competitiveness in OCR.
WeVisDoc-2B is an end-to-end document parser for page images, fine-tuned from Qwen3-VL-2B-Instruct. It converts pages into structured Markdown with LaTeX formulas and HTML tables. The model is available on Hugging Face under the Apache-2.0 license.
WeVisDoc-2B is a 2B-parameter model fine-tuned from Qwen3-VL-2B-Instruct. It achieves an Overall score of 95.06 on OmniDocBench v1.6 and a mean Overall score of 73.86 across the three PureDocBench tracks. The model card provides evaluation results comparing it with other end-to-end document parsing specialists, including WeVisDoc-4B which achieves 95.38 on OmniDocBench v1.6 and 75.54 mean on PureDocBench.
Developers outside China can now use a competitive open-weights document parser under Apache-2.0, reducing the cost of building document processing pipelines compared to proprietary OCR APIs.
The Apache-2.0 license allows commercial use, enabling businesses to integrate WeVisDoc-2B into document processing workflows without licensing fees.
Observable next signals include whether Tencent releases WeVisDoc-4B weights on Hugging Face and whether the model is adopted in downstream document processing tools.