A large language model, or LLM, is an artificial intelligence system trained to work with language. It learns patterns from large amounts of text and uses those patterns to predict and generate words. This lets it write answers, summarize documents, translate text, draft messages, explain code, and respond to questions.
An LLM does not understand language in the same way a person does. It uses mathematical patterns learned during training. When you give it a prompt, it estimates what text should come next based on the prompt and its training. Stronger models can follow instructions, use context, and produce useful explanations, but they can still make mistakes.
Systems like ChatGPT are built around language models. Some models run in cloud data centers, while smaller ones may run on personal devices with help from hardware such as a Neural Processing Unit.
LLMs are often combined with other tools. A system may use Retrieval-Augmented Generation to look up documents before answering, or it may call software tools to calculate, search, or update records. The model becomes the language interface, while other systems provide facts and actions.
The key benefit of an LLM is flexibility. The same model can help with many language tasks. The key risk is reliability: a fluent answer is not always a correct answer. Good use requires checking important information and designing systems that limit errors.