Coinbase Cuts AI Spending Nearly in Half by Defaulting to Open-Weight Models Like GLM and Kimi

CoinFeed June 29 news, Coinbase CEO Brian Armstrong shared an article on the company's experience in optimizing AI spending. He noted that while token usage has grown exponentially, through better default settings, routing, and caching strategies, AI spending has been cut nearly in half, rather than relying on usage caps and alerting mechanisms. Regarding defaults, Coinbase is defaulting to open-weight models (such as Zhipu's GLM 5.2 and Moonshot's Kimi 2.7) via its LLM gateway, while encouraging engineers to choose the right model for specific tasks. Ninety-one percent of the company's employees have never hit the usage cap, so the team is opting for cheaper defaults instead of lowering caps.