PulseAugur
EN
LIVE 11:21:10

WeChat integrates AI assistant Xiaowei powered by its 617B parameter WeLM model

WeChat has begun testing Xiaowei, an AI assistant integrated directly into its app. This assistant is powered by WeLM, WeChat's proprietary large language model, which has been developed with a sparse Mixture-of-Experts (MoE) architecture. A 617-billion-parameter version of WeLM activates only 23 billion parameters per token, and the development team is investigating "Hidden Decoding" as a method for further scaling. AI

IMPACT This integration signifies a trend of large consumer apps embedding advanced AI assistants directly into their platforms.

RANK_REASON This is a product integration of an existing model, not a frontier release from a major AI lab.

Read on Mastodon — fosstodon.org →

AI-generated summary · Google Gemini · from 2 sources. How we write summaries →

WeChat integrates AI assistant Xiaowei powered by its 617B parameter WeLM model

COVERAGE [2]

  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,…

  2. Mastodon — fosstodon.org TIER_1 English(EN) · [email protected] ·

    WeChat has begun testing Xiaowei, a native AI assistant embedded in the app, powered by WeLM. The model has grown to 617 billion parameters with a sparse MoE ar

    WeChat has begun testing Xiaowei, a native AI assistant embedded in the app, powered by WeLM. The model has grown to 617 billion parameters with a sparse MoE architecture, activating 23 billion per token. The team is exploring Hidden Decoding for new scaling directions. https:// …