Coinbase: Has reduced AI spending by nearly 50%, currently trying to default to open-weight models.

Deep Tide TechFlow news, June 27 - Coinbase CEO Brian Armstrong posted about the company's latest progress in AI cost optimization.

Armstrong said that as AI usage and token consumption continue to grow, the key to controlling costs is not limiting employee use or sending frequent budget reminders, but optimizing default model selection, task routing mechanisms, and caching strategies.

He revealed that Coinbase is using an internal LLM gateway to try open-weight models such as GLM 5.2 and Kimi 2.7 as default options, while still allowing engineers to select other models based on specific task requirements.

Data shows that 91% of the company's employees have never reached their AI usage limits, so Coinbase has not chosen to tighten quotas, but instead improves overall efficiency through lower-cost model solutions.

In terms of model routing, Coinbase preprocesses prompts and combines cache hit rates and pricing of different models to automatically assign tasks to the most suitable model.

Armstrong believes that complex tasks such as planning and reasoning may require cutting-edge model support, but execution-type tasks may not need to call more expensive models. In the future, the model selection process should be more automated by AI, rather than relying on human decision-making.

In addition, he pointed out that cache hit rate is one of the important factors affecting AI costs. Coinbase has incorporated cache-aware mechanisms into the request process to increase the reuse rate of historical results.

Taking LibreChat as an example, after optimizing the caching solution, its cache hit rate has increased from 5% to 60%.

Armstrong also said that the company expects engineers to keep context concise, including starting new sessions when switching tasks, narrowing file context scope, and turning off unused tools, to reduce unnecessary token consumption.

According to him, through the above measures, Coinbase has successfully reduced AI spending by nearly 50%, while token usage continues to grow.