Nathaniel Whittemore begins with reported AI policy discussions at the G7 and other industry headlines, distinguishing reports and commentary from settled outcomes.
The main segment looks at open-weight models and the tradeoffs among cost, local control, access and performanceAI inference economics examines the revenue, cost, utilization, pricing, and margins involved in serving trained models.. Whittemore describes a range of model comparisons and notes that benchmarks and user experiences can differ.
He discusses multi-model routingAI model routing sends each request to a model chosen for that request's complexity, cost, speed, privacy, or other requirements. and an example combining a lower-cost worker model with a frontier advisor. He argues that organizations will need to evaluate combinations of models against their own tasksEvaluation is the systematic assessment of a system's behaviour or results against explicit criteria. and budgets.
Watch on YouTube
