In 2022, the barrier to building with large language models (LLMs) was intelligence: reasoning quality, cost and performance. That gap has closed. In around two years, model costs have dropped ...