# Tencent releases WeVisOne-VL 4B vision-language model on Hugging Face

Digest AI · Generative AI & Models · published 2026-10-23T08:14:41Z

Canonical: https://digestai.news/story/tencent-releases-wevisone-vl-4b-vision-language-model-on-hugging-face

## Summary

Tencent’s WeChat Vision team has published WeVisOne-VL-4B, a unified vision‑language model that combines object‑level tasks such as open‑vocabulary detection, referring localization, counting, pointing and OCR with pixel‑level tasks like referring segmentation and promptable alpha matting. The model builds on Qwen3‑VL and uses structured next‑token prediction together with a lightweight 9.6 M‑parameter pixel decoder that handles bounding‑box and point coordinates via dedicated tokens.

The release includes an inference package on GitHub and a Hugging Face repository (tencent/WeVisOne-VL-4B). Users can load the model in a few lines of Python, specify a task (e.g., "matting"), and receive structured outputs such as masks and instance tokens. The code is under the Apache License 2.0, with third‑party components retaining their original licenses. The authors note that the model is intended for research in visual grounding and should not be used for autonomous high‑impact decisions or personal identification, and that fine alpha boundaries may be sensitive to image resolution and prompt phrasing.

## Key points

- Tencent's WeChat Vision released WeVisOne-VL-4B, a unified vision‑language model built on Qwen3‑VL.
- Model supports detection, localization, counting, OCR and matting via structured next‑token prediction and a 9.6 M‑parameter pixel decoder.
- WeVisOne-VL-4B is available on Hugging Face under Apache 2.0 for research use, with code and weights publicly released.

## Why it matters

An open‑source, multi‑task vision‑language model lowers the barrier for researchers and developers to build applications that need unified visual grounding capabilities.

## Sources

1. [tencent/WeVisOne-VL-4B · Hugging Face](https://huggingface.co/tencent/WeVisOne-VL-4B) (huggingface.co, 2026-10-23, primary source)

## Cite

Digest AI, "Tencent releases WeVisOne-VL 4B vision-language model on Hugging Face", 23 October 2026, https://digestai.news/story/tencent-releases-wevisone-vl-4b-vision-language-model-on-hugging-face

---

Written by Digest AI's editorial model from the linked sources; the sources are the record. Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse
JSON: https://digestai.news/story/tencent-releases-wevisone-vl-4b-vision-language-model-on-hugging-face.json