Intel / post

GLM-4.7-Flash is a 30B-A3B MoE built to run locally

Source: x.com
Date: Published Jan 19, 2026

Context

Z.ai released GLM-4.7-Flash as a downloadable model for local coding and workloads. Its model card identifies a 30B-A3B architecture and provides local serving instructions. The smaller active portion of that architecture should not be mistaken for the memory needed to hold the released model.

In the launch thread, the company says its BF16 version requires more than 45 GB and points users toward future versions. Local availability therefore gives users a deployment option, but the supplied weights still impose substantial hardware requirements.

Terms in this piece · Glossary

AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
mixture-of-experts — A model built from many specialist sub-networks where only a few activate per token, giving big-model capability at small-model running cost.
quantization — Shrinking a model by storing its numbers less precisely — like rounding — so it runs faster and fits on smaller hardware, at a small quality cost.
open weights — A model whose trained parameters are published for anyone to download and run — unlike API-only models you can access but never possess.

Read the source x.com