Back to today’s briefing

Story perspectives

Zhipu AI's GLM-4.6 Cuts Costs, Outshines Rivals

10/3/2025

28 5 Full Breakdown

Summary

Story summary

Zhipu AI launched GLM-4.6, cutting token use by 30% and costs by up to 85%.
It outperforms Chinese rivals in real-world coding and matches Claude Sonnet 4.5 on several tasks.
Anthropic's Claude Sonnet 4 faces AI-alignment challenges, though safety has improved.
Novo Nordisk and Cox Automotive report productivity gains from Claude models; GLM-4.6's efficiency strengthens its market position.

History

Historical Context

The introduction of Google's TensorFlow in 2016 marked a pivotal shift in AI research, enabling developers to cut computational costs by up to 90%. This trend of efficiency continued with Zhipu AI's GLM-4.6, which reduces token usage by 30% and costs by 85%. Such advancements highlight the competitive landscape in AI, where companies like Anthropic must innovate to tackle alignment challenges.

1 of 2

Story summary

Zhipu AI launched GLM-4.6, cutting token use by 30% and costs by up to 85%.
It outperforms Chinese rivals in real-world coding and matches Claude Sonnet 4.5 on several tasks.
Anthropic's Claude Sonnet 4 faces AI-alignment challenges, though safety has improved.
Novo Nordisk and Cox Automotive report productivity gains from Claude models; GLM-4.6's efficiency strengthens its market position.

1 / 2