Announcement dated August 26, 2026: Weixin says its 9B embedding model supports search and recommendations across text, images, and video, and led two evaluations.
In an announcement published on August 26, 2026, Weixin said its Vision team developed WeMM-Embedding to support search and recommendations across text, images, and video. The company says the 9B model is deployed in several of its services and ranked first in the MMEB-v2 and MMEB-v3 evaluations.
The announcement is relevant to teams assessing multimodal search: it describes an embedding model that, according to Weixin, is used in production. To verify the result, consult the company’s original announcement and check the scope and results of the MMEB-v2 and MMEB-v3 evaluations; the text provides no further details on methods or scores.