Small Language Models are the Future of Agentic AI

ABSTRACT

The rise of agentic AI systems has ushered in a mass of applications where language models perform specialized tasks repetitively with little variation. This position paper argues that Small Language Models (SLMs) are sufficiently powerful, inherently more suitable, and necessarily more economical for the vast majority of invocations within agentic systems. By challenging the industry's reliance on massive LLMs, the authors outline how SLMs can drastically reduce latency and infrastructure costs while maintaining high reliability in modular, multi-agent architectures.