AT&T has begun shifting significant portions of its artificial intelligence workloads away from Anthropic’s proprietary models toward open-source alternatives, achieving cost reductions of 80% to 90% on targeted applications. The telecom giant currently processes roughly 45 billion AI tokens per day, with 25% already running on open models, primarily for functions like network management. AT&T’s goal is to push that figure to between 70% and 80% over time, which has already saved millions of dollars. The company developed a proprietary « cache-aware AI Gateway » to route each query to the most cost-effective model capable of handling it efficiently.
Source: Read the original article
