F or the past few years, frontier AI has referred almost exclusively to the model layer – the largest, most capable systems produced by labs such as OpenAI, Anthropic and Google DeepMind, distinguished by scale, generality and the pace at which their capabilities have advanced. Each new generation has pushed further into reasoning, coding and autonomous agentic behaviour, closing the gap between what a model can theoretically do and what it can be trusted to do on an unsupervised basis.