Revolut moves to self-hosted GLM

AI summary

Editorial policy

Revolut moved about 2,000 engineers to self-hosted GLM for coding, moving Claude Code and Codex to an exception list.

Key points

•Approximately 2,000 engineers were moved to self-hosted GLM for coding tasks.
•The setup uses 56 Nvidia B200s with a blended cost of 30 cents per million tokens.
•This cost is roughly a tenth of what was previously paid to Anthropic and OpenAI.

Read the original

#AI Companies & Funding #Semiconductors & Infrastructure #LLM / Model Trends#Revolut#GLM#Nvidia B200

#AI Companies & Funding#Semiconductors & Infrastructure#LLM / Model Trends#Revolut#GLM#Nvidia B200

Hacker News9/18/2026

ZCode, the GLM coding agent, silently uploads your Git history(opens in a new tab)

The GLM coding agent ZCode has been reported to upload Git history without informing users.

Hacker News9/17/2026

GLM Built Its Own Inference Infrastructure

Z.ai explains the technical details of building its own inference infrastructure for GLM models to achieve inference efficiency and cost optimization.

Hacker News8/26/2026

Z.ai confirms Ox Alpha is a new GLM-series model and will release its weights(opens in a new tab)

中国のAI企業Z.aiは、DeepSeekと競合するステルスモデルOx AlphaがGLMシリーズに基づく新モデルであることを確認し、その重みの公開を予定している。

Hacker News8/25/2026

Ox-Alpha Is GLM?(opens in a new tab)

Hacker NewsでOx-AlphaというモデルがGLM（General Language Model）系なのかという議論が交わされている。技術的な詳細や関連性を探る投稿だ。

Cloud Watch11/6/2026

BoxWorks 2026: Box announces 'Box Context Retrieval' based on NVIDIA 'Nemotron'

Box announced 'Box Context Retrieval', developed based on NVIDIA's AI inference model 'Nemotron', revealing that it will be available starting in 2027.