[Europe's Top Fintech Giant Turns to GLM: AI Programming Bills Drop by Approximately 90%] Beating AI News reports that Revolut, Europe's highest-valued fintech unicorn, has adopted GLM as the default AI programming model for its approximately 2,000 engineers. Tools like Claude Code and Codex have been relegated to backups, primarily reserved for complex tasks. David Abitbol, Head of AI Operations at Revolut, revealed that the company previously allocated a $300 monthly budget per person for Claude Code, but this had limited impact on overall efficiency. In April of this year, the company began deploying GLM itself, offering engineers unlimited tokens. Currently, Revolut uses approximately 56 NVIDIA B200s to run GLM 5.3, translating to a total cost of approximately $0.30 per million tokens—only one-tenth of the costs incurred when using Anthropic and OpenAI. According to internal company statistics, engineers' coding time improved by approximately 60%, and their task completion time improved by approximately 70%. Previously, with limited tokens, engineers tended to reserve them for complex tasks. Now, without the need for careful token allocation, they are increasingly using AI to search internal documents, process Jira tickets, and locate requirements specifications in Slack. Only when tokens become cheap enough that they don't require meticulous management can AI have the opportunity to expand from a few complex tasks to the entire R&D process. [Original Link]