PulseAugur
EN
LIVE 09:32:59

WeChat's AI Assistant Xiaowei Runs on WeLM, a 617B Parameter MoE Model

By PulseAugur Editorial · [1 sources] ·

WeChat has begun testing Xiaowei, an AI assistant integrated within the app, which is powered by the company's large language model, WeLM. WeLM has been updated to a sparse Mixture-of-Experts (MoE) architecture, featuring a 617-billion-parameter version that utilizes 23 billion parameters per token. The development team is also investigating "Hidden Decoding" as a potential method for further scaling. AI

IMPACT This integration could significantly enhance user experience within WeChat by providing AI-powered assistance directly in the app.

RANK_REASON This is a product launch for an AI assistant integrated into an existing application, not a frontier model release.

Read on Pandaily →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

WeChat's AI Assistant Xiaowei Runs on WeLM, a 617B Parameter MoE Model

COVERAGE [1]

  1. Pandaily TIER_1 English(EN) · [email protected] (Pandaily) ·

    WeChat's Xiaowei Agent Runs on WeLM — a Sparse MoE Model Quietly Growing to 617 Billion Parameters

    WeChat has started gray-testing Xiaowei, a native AI assistant embedded inside the app, powered by WeLM, the WeChat AI team's long-developed large language model. WeLM has moved to sparse MoE architecture, with a 617-billion-parameter version activating just 23 billion per token,…

RELATED ENTITIES

RELATED TOPICS