Revolut Slashes AI Coding Costs 10x by Moving to Self-Hosted LLMs
British fintech company Revolut significantly reduced its AI coding expenses by tenfold by transitioning away from commercial APIs like Claude to self-hosted large language model solutions. This move highlights a growing industry trend where enterprises abandon costly commercial cloud APIs in favor of self-hosted open-source models to optimize software development budgets. Such cost-reduction strategies can heavily influence how engineering teams scale their internal AI tooling. By deploying open-source models on internal infrastructure, organizations gain tighter control over sensitive proprietary codebases while bypassing the steep per-token pricing charged by third-party model providers.
## BACKGROUND
Commercial APIs allow developers to integrate powerful AI capabilities into their workflows via cloud-hosted services, but usage costs can scale rapidly in large enterprises. In contrast, self-hosted LLMs involve running open-source AI models on an organization's own hardware or private cloud, providing greater privacy and predictability over operational expenses.