LLM

Large Language Model — a transformer-based model trained on internet-scale text to generate and reason.

Definition

An LLM (Large Language Model) is a transformer-based neural network trained on a large corpus of text to predict the next token. Modern frontier LLMs (GPT-5, Claude Opus 4, Gemini 2.5 Pro) are general-purpose: they can chat, reason, write code, and call tools. Their reasoning is statistical and unverifiable, so production systems combine them with tool use, retrieval, and human review.

When to use

Pick an LLM when the task involves natural language, fuzzy classification, summarization, or code generation. Avoid LLMs for tasks where a small classifier, regex, or rule engine would be deterministic and cheaper.

See also

Mentioned in