INSIGHT · US
AT&T cuts AI coding costs 56% with model routing and open-source tools
AT&T reduced coding and advanced AI task costs by 56% using model routers to direct queries to cheaper open-source models where appropriate, with only 2% quality degradation. Proof that cost optimization architecture is already shipping at scale.
WHY IT MATTERS
Demonstrates that enterprises can realize massive AI productivity gains without paying for premium frontier models for all workloads. BFSI should apply same logic to risk, compliance, and data extraction tasks.
Source: PYMNTS · 2026-08-20