Definition
What is large language model?
A large language model (LLM) is a model trained on very large text corpora to predict continuations of text. In practice this makes it able to classify, extract, summarise, translate and draft, given instructions and context, without task-specific programming.
Updated 3 September 2026
Why it matters
- One capability covers many previously separate tools — classification, extraction, summarisation, drafting.
- Behaviour depends heavily on the context supplied, which makes retrieval and prompt design engineering work.
- Model versions change; anything in production needs an evaluation suite to detect regressions.
Business example
A service team uses one model with different instructions and retrieved context to classify inbound tickets, summarise long threads and draft replies, replacing three separate tools.
Common misconceptions
- 'The model knows our business.' It knows what you put in its context window plus general training data.
- 'Bigger models are always better.' Smaller models are often sufficient and materially cheaper at volume.
- 'Output is deterministic.' Repeated calls can vary; evaluation must be statistical, not anecdotal.
Related concepts
- Generative AI
- Retrieval-augmented generation
- AI agent
Where would this apply in your business?
Merjora maps your workflows and quantifies the opportunities worth acting on.
Discover your AI opportunitiesRelated reading
Editorial standard. Merjora publishes analysis, frameworks and publicly documented examples. We do not publish invented statistics, unattributed benchmarks or unverified customer stories.