Quick answer
A large language model is a neural network trained on huge amounts of text to generate and understand language. GPT-4, Claude and Llama are examples.
What it is
A large language model (LLM) is a deep-learning model trained on massive text corpora. It learns patterns in language and can generate text, answer questions, summarise and reason. Modern LLMs are also multimodal, working with images, audio and code.
Why it matters
LLMs power most modern AI products: chatbots, copilots, search and content tools. Understanding their strengths and limits is essential to using them safely.
How to use it
- Pick a model based on the task: reasoning, summarisation, code, vision.
- Use system prompts to set behaviour and constraints.
- Ground the model in your own data with retrieval (RAG) when accuracy matters.
- Always have a human in the loop for high-stakes outputs.
Examples
- A chatbot uses an LLM to answer questions grounded in your product docs via RAG.
- A legal team uses an LLM to summarise contracts, with a lawyer reviewing every output.