Revolut cuts AI coding costs with self-hosted open models
Revolut says it cut its AI coding inference costs by about 90% by making self-hosted open-weight models its default choice.
By GennnAi··1 min read
David Abitbol, Revolut's operations and infrastructure lead, described the shift at the AI Engineer conference. Some 2,000 engineers now use Z.ai's open-weight GLM models running on 56 Nvidia B200 GPUs. Anthropic's Claude Code and OpenAI's Codex are not banned: they remain available as exceptions to the default. The company says the cost drop was achieved in under a year, and the productivity figures cited come from its own presentation.
Source: The Register, published on November 3, 2026.