Small Language Model

Small Language Models (SLMs) are lightweight, highly efficient neural networks designed to deliver fast, specialized natural language processing directly on local devices without cloud dependency.

While massive frontier models capture headlines, Small Language Models (SLMs) handle targeted production tasks at a fraction of the cost. These models typically operate with fewer than 10 billion parameters (popular examples include Microsoft's Phi-3 and Google's Gemma 2B) and run comfortably on consumer-grade hardware like smartphones and laptops. By trading broad, generalist knowledge for narrow domain expertise, SLMs eliminate high cloud API fees, dramatically reduce latency, and keep sensitive data fully secure on-device. They represent a practical shift in AI deployment: choosing lean, specialized efficiency over raw, expensive scale.

https://huggingface.co/blog/introduction-to-small-language-models