Revolut moves to self-hosted GLM
105 points by pretext
Revolut has transitioned approximately 2,000 engineers to a self-hosted General Language Model (GLM) for coding, while excluding Claude Code and Codex from regular use. According to David Abitbol, the head of AI ops at Revolut, the company is utilizing 56 Nvidia B200 servers at a cost of 30 cents per million tokens, significantly reducing expenses compared to previous payments to Anthropic and OpenAI.