Revolut sidelined Claude for a Chinese model, cut coding bill 90%

Revolut moved its roughly 2,000 engineers off Anthropic's Claude Code and OpenAI's Codex as their default coding tools and onto GLM, an open-weight model from Chinese lab Z.ai that the bank hosts itself, and says its coding bill is now about a tenth of what it paid the American labs.

The numbers come from Revolut's AI operations lead David Abitbol, speaking at the AI Engineer conference in New York: about $0.30 per million tokens on 56 Nvidia B200 GPUs, with coding time down 60 percent and lead time down 70 percent after the rollout, the company's own figures, not an independent audit.

Claude and Codex are not gone, just demoted to the exception for the hardest work and vision tasks, which is the point: when tokens are nearly free, engineers stop rationing them and spend them on debugging and ticket drudgery instead, and the frontier labs' pricing looks like the expensive habit.