Large language model (LLM)
What is a large language model?
A large language model (LLM) is an AI model trained on very large amounts of text to predict the next word (more precisely, the next token) in a sequence. That simple training goal lets it write, summarise, translate, answer questions and follow instructions. GPT models behind ChatGPT, Google’s Gemini, Anthropic’s Claude and Meta’s Llama are examples.
An LLM’s built-in knowledge is frozen at its training cut-off and stored as patterns, not as a database of facts with sources. It can be wrong while sounding confident, which is often called hallucination. To answer with current, sourced information, assistants combine the model with search, a technique called retrieval-augmented generation.
Most LLMs today are built on the transformer architecture, introduced by Google researchers in 2017. Google also uses language models inside Search to understand queries and content, for example BERT from 2019 and MUM from 2021.
Why large language model matters for SEO
LLMs now sit between many searchers and the web. When they write an answer, they decide which brands to name and which pages to cite, so understanding how they work, and where their knowledge comes from, is the basis of LLM SEO and AI visibility.
They are also a content tool. Many teams use LLMs to draft, outline or edit. Google allows that, as long as the result is helpful and accurate; see AI-generated content.
Large language model FAQ
Do LLMs search the web?
The model on its own doesn’t. Assistants built on LLMs, such as ChatGPT, Gemini, Claude and Perplexity, can call a search tool and give the results to the model, which is how they cite current sources.
Why do LLMs make things up?
An LLM generates likely text rather than looking up facts, so it can produce plausible but false statements. Grounding the answer in retrieved sources reduces this but doesn’t remove it.
Related SEO terms
Retrieval-augmented generation (RAG)
Retrieval-augmented generation (RAG) is when an AI system fetches relevant documents first and then writes its answer from them.
LLM SEO is optimising your content and brand so large language models such as ChatGPT, Gemini and Claude mention and cite you in their answers.
Grounding is tying an AI model’s answer to real, retrievable sources, such as search results, so it is current and can be checked.
Semantic search is search that matches the meaning and intent behind a query, not just its exact words.
AI-generated content is text, images or other media produced with generative AI tools. Google judges it by quality, not by how it was made.