Revolut Cuts AI Coding Inference Costs by 90% Using GLM

3 Nov 2026, 5:07 pm · 3h ago · 1 min read · Analytics India MagazineNew

British fintech major Revolut has reduced its AI coding inference costs by roughly 90% compared to its earlier spending on OpenAI and Anthropic models. The digital bank achieved this by self-hosting the open-weight GLM model from Chinese AI lab Z.ai on dedicated Nvidia GPUs. Speaking at the AI Engineering conference, Revolut AI operations lead David Abitbol stated that blending costs dropped to 30 cents per million tokens, while improving engineering lead times by 70% and coding time by 60% across 2,000 engineers. Revolut was valued at $115 billion in July.