Small Language Models (SLMs) Can Still Pack a Punch: A survey (updated 2026)

TL;DR

This survey reviews small language models (1-8 billion parameters) showing they can match or outperform larger models, emphasizing efficiency and scalability in AI development.

Contribution

It introduces a comprehensive overview of SLMs, their capabilities, and techniques for balancing performance with cost, guiding future model development.

Findings

SLMs can perform as well or better than larger models.

Techniques exist to optimize SLMs for efficiency and scalability.

SLMs' effective sizes reflect increased capabilities relative to LLMs.

Abstract

As foundation AI models continue to increase in size, an important question arises - is massive scale the only path forward? This survey of about 160 papers presents a family of Small Language Models (SLMs) in the 1 to 8 billion parameter range that demonstrate smaller models can perform as well, or even outperform large models. We explore task agnostic, general purpose SLMs, task-specific SLMs and techniques to create SLMs that can guide the community to build models while balancing performance, efficiency, scalability and cost. Furthermore we define and characterize SLMs' effective sizes, representing increased capability with respect to LLMs.