Large Language Model (LLM)
What is Large Language Model (LLM)?
A large language model (LLM) is an AI model trained on very large amounts of text to understand and generate language. It powers most chat assistants, writing tools and generative AI features.
An LLM works by predicting the most likely next piece of text (a token) given everything that came before it, which lets it answer questions, summarize, translate, classify and draft. The largest models are general purpose: they know a little about almost everything, which makes them flexible but expensive to run at contact-center volumes and harder to keep within policy.
For customer service, the questions that matter are where the model runs, what data it sees, whether that data is used to train it, how its answers are grounded in your own knowledge, and what it costs per interaction. Many teams now combine a private LLM for open-ended tasks with smaller, specialized models for high-volume, well-defined ones.
Related: Small language model (SLM), AI models (SLMs and LLMs), Tokens and context window.