BlockBeats News, June 28th, Coinbase CEO Brian Armstrong revealed that the company has reduced AI expenses by nearly half by switching default AI models to open-source models such as GLM 5.2 and Kimi 2.7, implementing intelligent model routing, cache optimization, and other measures, while experiencing an exponential increase in Token usage.

Brian Armstrong stated that Coinbase no longer relies on usage limits or spending notifications to control costs. Instead, the company adopts cost-effective open-source models by default and only calls upon more powerful cutting-edge models in necessary scenarios. Additionally, the company further reduces inference costs through automatic model routing, increased cache hit rate, and streamlined context.

It is worth noting that this statement comes at a time when Anthropic's CEO has recently expressed concerns about open-source models and model distillation. Coinbase's practice demonstrates that in enterprise-level AI applications, open-source models are becoming an important choice to balance cost and efficiency.