British fintech Revolut slashes AI coding costs 10x by cutting out Claude and going self-hosted
British fintech Revolut achieved a 90% reduction in AI coding inference costs by deploying self-hosted GLM models using 56 Nvidia B200 GPUs instead of relying on commercial services like OpenAI and Anthropic. The company reported a 60% improvement in coding time and 70% improvement in lead time after rolling out AI coding tools to 2,000 engineers. Self-hosting GLM 5.1 costs approximately $0.30 per million tokens compared to significantly higher rates for Claude Code and OpenAI's Codex.
Key points
- Self-hosted GLM models cost ~90% less than commercial AI services
- Coding speed improved 60%, lead time improved 70%
- Uses 56 Nvidia B200 GPUs for inference at $0.30 per million tokens
Summary and key points written by AI from the headline and a short excerpt, so they can be incomplete or wrong. The Register did the reporting; read the original for the full story.
Read the full story at The Register
