Large language models (LLMs) such as GPT-4o and other modern state-of-the-art generative models like Anthropic’s Claude, Google’s PaLM and Meta’s Llama have been dominating the AI field recently. These models have enabled advanced NLP tasks such as high-quality text generation, answering complex questions, code generation and even logical reasoning. At the same time, these gigantic models are resource-hungry and are hindered by their size and complexity. LLMs require significant amounts of computing power and infrastructure. Think of your smartphone, your smart TV or even your fitness tracker—these devices don’t have the computational power to run large models like LLMs effectively. Small language models (SLMs) are lightweight neural network models designed to perform specialized natural language processing tasks with fewer computational resources and parameters, typically ranging from a few million to several billion parameters. Unlike large language models (LLMs), which aim for general-purpose capabilities across a wide range of applications, SLMs are optimized for efficiency, making them ideal for deployment in resource-constrained environments such as mobile devices, wearables and edge computing systems.

Full commentary : How small language models could power the edge devices of future.