Unverified60% confidenceFactTime unknown

GLM 4.6相较于GLM 4.5，代码能力提升27%，token消耗降低30%

2

Sources

60%

Confidence

Medium-term (~90 days)

Relevance

6/3/2026

First Seen

Valid until: 9/1/2026

Sources

Claude Code搭配GLM4.6：编程Agent国产化部署指南

bilibili潜龙悟用-

Related Entities

GLM 4.6model GLM 4.5model

Related Claims

UnverifiedGLM-5.2 在 Frontiers-WE 基准上从上一代的30出头提升到7469% similar UnverifiedGLM-5.2 在 Deeps-WE 基准上从上一代的18提升到4665% similar UnverifiedGLM 5.2是一个开源高性能模型，据称能达到接近甚至超越Opus 4.8的效果，价格却只有六分之一65% similar UnverifiedGLM5的上下文长度超过20万token时会出现明显的上下文腐烂问题65% similar UnverifiedGLM5.2的推理速度明显慢于同级别模型，是其最大短板64% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/38491

API

curl https://kongchang.com/api/v1/knowledge/claims/38491

MCP

get_claim(id=38491)