Large language model (LLM)

A large language model (LLM) is an AI ✕ Artificial intelligence is a computer system’s ability to perform tasks we associate with human intelligence, such as understanding language, recognizing images or making decisions. More in the glossary → system trained on huge amounts of text to understand and generate human language. At its core it works by predicting, word by word (or token ✕ A token is the smallest unit of text a language model works with: usually a word or a piece of a word. Pricing and limits are measured in tokens. More in the glossary → by token), the most likely continuation of a text; by repeating that prediction it produces answers, summaries, translations or code. It is the technology behind assistants like ChatGPT ✕ GPT is the family of large language models from OpenAI behind ChatGPT. It generates text by predicting, unit by unit, the most likely continuation. More in the glossary → , Claude or Gemini. Its great strength is versatility, but it also has clear limits: it can invent facts with complete confidence ( hallucinations ✕ A hallucination is when an AI presents made-up or incorrect information as true, in a confident tone. It is why you should verify its answers. More in the glossary → ), it does not tell true from false on its own, and its knowledge is limited to what it saw during training. That is why it is best to treat its answers as a useful draft that needs checking, not as an infallible source.