The systematic process of evaluating and choosing the most appropriate AI model for a specific task based on performance, cost, and latency requirements.
Overview
- [Emerging signal] Enterprises should implement a ‘model sommelier’ role to continuously evaluate and select the most cost-effective models for specific tasks, rather than defaulting to the most advanced frontier models. (Source: AI Daily Brief Host, via AI Daily Brief, 2026-08-24)