Tencent Open-Sources WeVisOne-VL for Unified Visual Perception

Tencent's WeChat Vision team open-sourced WeVisOne-VL, a vision-language model built on Qwen3-VL (2B/4B), which unifies object-level and pixel-level perception in a single model supporting detection, counting, OCR, segmentation and matting.

2026-10-23 ~ 2026-10-23 · 3 related posts

• Tencent's WeVisOne-VL-2B model card lands on Hugging Face — tencent · 2026-10-22
• Tencent's WeChat Vision releases WeVisOne-VL, unifying object- and pixel-level perception — jacek2023 · 2026-10-23

1 near-duplicate retellings: tencent