A large language model (LLM) is a type of artificial intelligence trained on an enormous amount of text data that can understand natural language and generate text – answering questions, writing, summarising, translating or coding. Its technological foundation is the transformer architecture: during training, the model learns statistical relationships between words split into tokens, and when generating, it always predicts the most probable next word in context. The best-known LLMs include the GPT models from OpenAI, Claude from Anthropic, Gemini from Google, and open models such as Llama or Mistral. The model's capabilities are determined by the size and quality of the training data, the number of parameters, and the size of the context window, that is, the amount of text the model processes at once. LLMs also have limits: they do not know events after their training date, they can hallucinate, meaning generate convincing-sounding falsehoods, and their output depends on the quality of the prompt. In companies, LLMs are deployed through APIs, chatbots, RAG systems and AI agents.
See also: AI Hallucination, Context window.
