A large language model (LLM) calculates how likely a next piece of text is. Google explains in its course on large language models that a language model estimates the probability of a token occurring within a longer sequence. A token can be a word, part of a word or a single character. An LLM is such a model at large scale.
Anthropic, for example, describes Claude as a family of large language models in its models overview.
Why does it matter?
A language model does not look up facts in a database. It knows what was in its training data, up to its knowledge cutoff. Newer or specific information, such as your current prices or opening hours, it only knows if the AI assistant searches live and finds your page. That is why a readable website and accurate business information count.
Example
Suppose you type “The best time to sow a lawn is”. The model picks the most likely continuation, step by step. If that continuation sounds plausible but is wrong, it is called a hallucination.
Common mistake
Treating every system with “AI” in its description as a language model. Atlas, the research engine developed by Outseen, is an engine and not an AI model: its technical core works without a language model.
More in How do AI assistants decide which websites to mention?.