Large Language Model

LLM means Large Language Model.

• Large — scale in parameters, training compute, and training data.
• Language model — learns a probability distribution over sequences of tokens.

Most modern LLMs are pretrained on huge text corpora, then further trained so they behave more like helpful assistants (instruction following, safer defaults, better conversation habits).

Put the pieces together: an LLM is a large Transformer-based (often decoder-only) model that predicts likely next tokens — and after extra training, those predictions look like answers, plans, and code.