What is a large language model (LLM)?
An LLM is a model trained on huge amounts of text or similar data in order to work with language.
What is it?
A large language model (LLM) has been trained on huge amounts of text and can predict what comes next in a text. From that come its abilities to write, summarise, translate and answer. Chatbots such as ChatGPT, Claude and Gemini are built on such models.
An everyday example
When a chatbot sums up a long email thread in three bullet points, a large language model is doing the work.
Why does it matter?
LLMs have become a general tool for working with text and code. Understanding how they work makes it easier to judge when to trust them.
A common misconception
That fluent writing means checked facts. The model writes likely text, which is not the same as true text.
In more detail
How does it build a sentence?
Based on the context, it chooses the next token, then repeats the process. This allows surprisingly complex output.
Does it “understand” language?
Models show many abilities that look like understanding. Whether that is the same as human understanding is a different, open question.
Why does it work for so many tasks?
Because the patterns in huge amounts of data are useful for writing, summarising, translating, programming and reasoning.
Why does it make things up?
Because language that sounds likely is not automatically a checked claim. Extra tools and checking can help, but they do not guarantee complete accuracy.
Remember: Fluent language is not proof that the model checked the facts.
Reliable sources
Last reviewed 1 November 2026. An explainer, not professional, legal or medical advice. For rules and figures that change, check the official version.