Home
WeChat WeMM-Embedding Powers Moments Search and Video Highlights With 1 Billion Daily Calls
On September 4, WeChat AI announced the open-source release of its general multimodal embedding model, WeMM-Embedding. Available in 2B, 4B, and 9B versions, the model supports text, images, videos, visual documents, and any interleaved multimodal input.
According to WeChat employee @KeCunXiaoJiang, the WeChat Visual Team’s open-sourced multimodal model, WeMM-Embedding, enables the system to simultaneously understand text, images, and videos by placing them in a shared semantic space for comparison and retrieval.

Already applied in scenarios such as Moments search and video recommendation
He revealed that this model is widely used within WeChat—Moments search, video recommendation, public account recommendation, and WeChat e-commerce all use it, with a daily call volume of 1 billion times. On the international authoritative benchmark MMEB-v2, the model ranked first, surpassing all previously submitted open-source and closed-source models.
Currently, related resources of WeMM-Embedding have been made available: the paper was published on arXiv, the code repository is hosted under the Tencent/WeMM-Embedding project on GitHub, and the model set is also available on Hugging Face.
Related article
Anthropic Hits $1.5B Copyright Deal to Compensate 500,000 Works
On Monday, a U.S. federal court granted final approval for a $1.5 billion class-action settlement between Anthropic and a coalition of authors and publishers, as reported by Reuters. The deal resolves ongoing litigation regarding Anthropic’s alleged
Xiaohongshu, Peking University, Shanghai Jiao Tong University Unveil HYPIC to Slash First Token Latency by 3.25x in Hybrid Attention Models
The primary focus of large model services is transitioning from single-turn conversations to retrieval-augmented question answering, multi-document summarization, and long-range agents. These emerging workloads share a key trait: a single request's p
ChatGPT Unveils Custom Stickers Seamlessly Compatible With Top Social Media Platforms
OpenAI has officially introduced a new custom sticker feature for ChatGPT, pushing the boundaries of AI in mass social content creation.Users can create personalized stickers through intuitive methods, including uploading photos, using text prompts,
Related Special Topic Recommendations
Comments (0)
0/500
On September 4, WeChat AI announced the open-source release of its general multimodal embedding model, WeMM-Embedding. Available in 2B, 4B, and 9B versions, the model supports text, images, videos, visual documents, and any interleaved multimodal input.
According to WeChat employee @KeCunXiaoJiang, the WeChat Visual Team’s open-sourced multimodal model, WeMM-Embedding, enables the system to simultaneously understand text, images, and videos by placing them in a shared semantic space for comparison and retrieval.

Already applied in scenarios such as Moments search and video recommendation
He revealed that this model is widely used within WeChat—Moments search, video recommendation, public account recommendation, and WeChat e-commerce all use it, with a daily call volume of 1 billion times. On the international authoritative benchmark MMEB-v2, the model ranked first, surpassing all previously submitted open-source and closed-source models.
Currently, related resources of WeMM-Embedding have been made available: the paper was published on arXiv, the code repository is hosted under the Tencent/WeMM-Embedding project on GitHub, and the model set is also available on Hugging Face.
Anthropic Hits $1.5B Copyright Deal to Compensate 500,000 Works
On Monday, a U.S. federal court granted final approval for a $1.5 billion class-action settlement between Anthropic and a coalition of authors and publishers, as reported by Reuters. The deal resolves ongoing litigation regarding Anthropic’s alleged
Xiaohongshu, Peking University, Shanghai Jiao Tong University Unveil HYPIC to Slash First Token Latency by 3.25x in Hybrid Attention Models
The primary focus of large model services is transitioning from single-turn conversations to retrieval-augmented question answering, multi-document summarization, and long-range agents. These emerging workloads share a key trait: a single request's p
ChatGPT Unveils Custom Stickers Seamlessly Compatible With Top Social Media Platforms
OpenAI has officially introduced a new custom sticker feature for ChatGPT, pushing the boundaries of AI in mass social content creation.Users can create personalized stickers through intuitive methods, including uploading photos, using text prompts,











