The economic efficiency of an AI model or system, evaluated by the ratio of output quality or task completion rate to the total cost of inference and deployment.
Overview
- A study by Alpha Sense found that GPT-5.6 Soul completed financial analysis tasks for approximately 13% cheaper than Kimi K3 while achieving a quality score around 20% higher. (Source: Alpha Sense study cited by AI Daily Brief host, via AI Daily Brief, 2026-08-24)