Small language models (SLMs) specialize in GenAI operations, having a smaller blast radius, and use less compute that their larger siblings, LLMs.