Large language model

A large language model, or LLM, is a parameter-rich model trained on language sequences to predict and generate tokens in context.

Plan for this page

Understand one concept in plain language and connect it to nearby ideas.

You will leave with: LLMs learn statistical regularities that support drafting, transformation, classification, coding, question answering, and other tasks. A deployed assistant may add instructions, retrieval, tools, memory, filters, and interfaces beyond the LLM itself.
Time: 3–8 min
Before you begin: No prerequisite. Keep one real example in mind.
Do this now: Read the explanation, then compare the example with the common misunderstanding.

Prepared next stepExplore tokens, context, and prediction

Large language modelA large language model, or LLM, is a parameter-rich model trained on language sequences to predict and generate tokens in context.

Foundation model
Token
Context window
Prompt

Plain explanation

LLMs learn statistical regularities that support drafting, transformation, classification, coding, question answering, and other tasks. A deployed assistant may add instructions, retrieval, tools, memory, filters, and interfaces beyond the LLM itself.

Example

Given a paragraph and the instruction to summarize it, an LLM generates a shorter token sequence conditioned on both.

Why it matters

LLMs can produce fluent and flexible language without guaranteeing factual support or human-like understanding. Separating model from product reveals where context, evidence, tools, and controls enter.

Common misunderstanding

Misconception: an LLM stores a verified encyclopedia and looks up the correct sentence. Correction: it generates likely token sequences from context and learned parameters.

Related concepts

Follow relationships, not only the alphabet

Foundation modelA foundation model is trained broadly enough to be adapted or used across multiple downstream tasks and applications.TokenA token is a unit into which a model’s input or output is divided for processing.Context windowA context window is the limited amount of tokenized information a model can consider together during one generation or inference.PromptA prompt is input supplied to a generative model or system to shape the task, context, constraints, or desired output.

Explore tokens, context, and prediction

Sources and scope

What supports this object

Machine Learning GlossaryGoogle for Developers

Defines large language models and related terms for technical learners; model families, training mixtures and product wrappers differ.

Open source

Loading page content...

…