AI Glossary
LLM (large language model)
Large Language Model, large language model
An LLM is a large language model trained on vast amounts of text that predicts the next token and, on that basis, generates answers, summaries or code in natural language.
- Learns the statistical patterns of language from large bodies of text.
- Works by predicting the next token, not by understanding facts.
- Quality depends on the data, the number of parameters and how it has been tuned.
An LLM is a machine-learning model trained on very large amounts of text. During training it adjusts billions of parameters so as to predict the next token in a sequence as accurately as possible. From this simple principle comes its ability to write, translate, summarize and answer questions.
It is worth remembering that the model works on probability, not on verified knowledge. That is why it can sound confident and still be wrong — a phenomenon we call hallucination. In business applications an LLM is often paired with document retrieval so that answers are grounded in specific sources.
Related terms
In guides
Related articles