What it means
A Large Language Model is a type of artificial intelligence trained to understand and generate text. The model learns patterns from billions of words during training, then uses those patterns to predict the most likely next word in a sequence. That simple mechanism, applied at scale, produces conversation, summarization, translation, and answer generation.
Modern LLMs include OpenAI's GPT-4 and GPT-5 (powering ChatGPT), Anthropic's Claude family, Google's Gemini, Meta's Llama, and Mistral.
Why it matters
Every AI search engine in 2026 sits on top of an LLM. When ChatGPT recommends a Melbourne plumber, the LLM is what synthesizes the answer. Understanding how LLMs decide which businesses to cite is the foundation of Answer Engine Optimization.
LLMs cite sources for two reasons: the source appeared frequently and authoritatively in training data, or the source ranked highly when the LLM searched the web in real time. Both are addressable through SEO and AEO work.
How it's used
For business optimisation, the relevant levers are:
- Training data presence - get mentioned in trustworthy web sources before a model's training cutoff
- Live retrieval optimization - rank well in Google and Bing because LLMs pull from those results
- Brand consistency - use the same business name across all sources so the model associates the right entity
- Answer-shaped content - write paragraphs the LLM can lift cleanly
For depth, see our GEO SEO complete guide.