CoolingOpen source
Revolut says self-hosting GLM models cut its AI coding costs about 90%
Revolut ops lead David Abitbol told the AI Engineer conference the bank cut AI coding inference costs roughly 90% by making self-hosted Z.ai GLM models its default, running GLM-5.3 on 56 Nvidia B200s for about 2,000 engineers, per The Register. He cited $0.30 per million tokens and 60% faster coding time. The figures are the company's own and unverified.
- HEAT
- OUTLETS
- 2
- FIRST SEEN
- LAST SIGNAL
Primary source
The Registertheregister.com/ai-and-ml/2026/11/03/british-fintech-revolut-slashes-ai-coding-costs-10x-by-cutting-out-claude-and-going-self-hosted/5307345Coverage · 2 articles, oldest first
Tuesday, November 3, 2026