Large Language Model (LLM)

A large language model, or LLM, is a neural language model trained on large collections of text so it can model patterns in language and perform many tasks through prompting. Many modern LLMs are based on the Transformer architecture and are trained before they are adapted to specific products or tasks.

During pre-training, the model learns to predict or reconstruct language from large datasets. Afterward it can be adapted through instruction tuning, Fine-tuning, preference optimization, retrieval, or tool use. Scale alone does not define an LLM, but model capacity, training data, compute, and broad task coverage are central characteristics.

LLMs are the foundation of many conversational, coding, search, and agentic systems. They can generate fluent answers, but fluency does not guarantee factual accuracy, which is why grounding, evaluation, and hallucination controls matter.

See also GPT, Foundation Model, Retrieval-Augmented Generation (RAG), and Context Window.