AI Briefing
KO

Tencent Releases WeMM-Embedding

·2026.08.25 18:36

Key point

Tencent has released WeMM-Embedding, a general-purpose multimodal embedding model based on Qwen3.5.

Details

Tencent has released the WeMM-Embedding series, a general-purpose multimodal embedding model based on Qwen3.5.

The model supports text, image, video, visual document, and cross-modal inputs, generating 4,096-dimensional L2-normalized embeddings. Audio input is not supported.

The model is available in three sizes: 9B, 4B, and 2B. The technical report and model weights can be found on Hugging Face and GitHub.

open-sourcetencentqwen-3.5ai-modelmultimodal-embedding
Read originalShare on Threads

This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.

Related

RedditOpen-Weight CLM Released as Jev Alternative, Agent Decision-Making 4x FasterPyTorchKRTrillion Labs Releases GravityOCR, an OCR Model Applying Self-Speculative DecodingHacker NewsJevper Released: A Jev Classification Wrapper for OpenAI-Compatible Models with Automatic Logprobs DetectionGeekNewsRootless Linux Sandbox 'Drop' Released for Isolating AI Agents and Third-Party ProgramsPyTorchKRPostgreSQL Extension pgjev Released, Supporting Natural Language Filtering Without EmbeddingsTLDR AIJared Palmer Releases 'Kev', a Lightweight Decision-Making Model Based on Qwen3.5