CompaniesOtherNews brief

Tencent WeChat Open-Sources WeMM-Embedding Multimodal Models

Published: Updated: By 24TopNews Editorial Desk

Tencent's WeChat vision team has open-sourced the WeMM-Embedding series of multimodal embedding models, which handle text, images, and video. The models, available in 2B, 4B, and 9B parameter versions, are already deployed across WeChat Channels, Official Accounts, Moments, and e-commerce services. Tencent's evaluations show the 9B version ranks first on the MMEB-v2 and MMEB-v3 benchmarks.

Tencent's WeChat vision team has announced the open-sourcing of the WeMM-Embedding series of multimodal embedding models, which can represent and match various content types including text, images, and video. The team said these models have already been deployed in WeChat Channels, Official Accounts, Moments, and e-commerce services. The release includes three parameter versions: 2B, 4B, and 9B.

According to Tencent's evaluation results, the 9B version ranks first among the listed models on both the MMEB-v2 and MMEB-v3 benchmarks.

24TOPNEWS IMPACT INTELLIGENCE

Why this event matters

The event has a measured impact on 2 industrys. The strongest current signal is positive for Artificial Intelligence, with intensity 70/100 and 80% confidence over a short term horizon.

Technology · 10.4

Artificial Intelligence

Direction
positive
Intensity
70
Confidence
80%
Horizon
Short term
Effective impact +39
Internet & Media · 11.6

Diversified Internet Platforms

Direction
positive
Intensity
60
Confidence
75%
Horizon
Short term
Effective impact +32

Impact figures are analytical estimates that combine direction, intensity, confidence and event importance. They are not investment advice.