Theo Browne examines a reported doubling of usage limits on several OpenAI subscription tiers in the context of earlier changes. He compares subscription allowances and discusses how limits and resets affect the value of coding models.
Browne distinguishes token pricesAI inference economics examines the revenue, cost, utilization, pricing, and margins involved in serving trained models. from the cost of completing a taskCost per completed task measures the total AI, tool, infrastructure, retry, and repair expense for each verified useful outcome., then describes his own comparison of model-generated code reviewsEvaluation is the systematic assessment of a system's behaviour or results against explicit criteria.. His preferences and estimates reflect the conditions and data available when the video was recorded.
Watch on YouTube
