Glossary

What is a large language model (LLM)?

A large language model, or LLM, is an AI system trained on huge amounts of text so that it can understand and write language. It builds a reply by predicting the most likely next word, over and over. That lets it draft, summarise, translate and answer questions. GPT, Claude, Gemini, Llama and DeepSeek are all large language models.

Why it matters for a business

An LLM produces fluent text from patterns, not from a store of checked facts. That is why it is excellent at drafts and rewrites, and why it can state something false in the same confident voice as something true. On its own it cannot look anything up, act on a system or remember past conversations. Products add those around it, and most of what makes one AI product better than another sits in that surrounding layer rather than in the model.