According to CoinMeta, Tencent WeChat's visual team has open-sourced Wemm-Embedding based on Qwen3.5, and launched three multimodal Embedding models: 2B, 4B, and 9B. These models are capable of converting text, images, videos, and visual documents into vectors for use in search, recommendation, and content matching. The 2B model achieved a score of 77.9 on MMeB-V2, slightly higher than the 8B model's score of 77.8, while the 9B version further improved to 80.6 points. Tencent stated that these models have been practically applied in WeChat official accounts, video accounts, e-commerce recommendations, and WeChat search.