What is an LLM?
A plain-language introduction to large language models and how they work.
A large language model (LLM) is a neural network trained on enormous amounts of text to predict the next token in a sequence. Modern LLMs like GPT-5, Claude Sonnet 4.5 and Gemini 2.5 Pro use the transformer architecture introduced in 2017 and are trained on trillions of tokens. After pretraining, they are typically fine-tuned with instruction data and reinforcement learning from human feedback (RLHF) so they follow instructions helpfully. This guide walks through tokens, context windows and why LLMs are good at some tasks and unreliable at others.