AI glossary: short definitions of key AI terms
Short, plain definitions of AI terms, with links between related entries.
AI texts are full of terms that sound alike and mean different things. This glossary gives each term a short, plain definition.
How to read it
- First line first. Each entry opens with a one-sentence definition. That sentence is enough if you only need the gist.
- Then the detail. The rest explains why the term matters for cost, quality or safety.
- Follow the links. Entries link to related terms. Start with token, context window, prompt or hallucination and branch out.
What the glossary does not do
It does not rank products or tell you what to buy. Where an entry states a number, it names the source and the date we checked it.
Missing a term?
Tell us in the form at the bottom of any page. The list grows with each update round. The full list of terms follows below.
All pages in this section (30 terms)
A
- AI agent
An AI agent is a system where a model plans, uses tools and checks results in a loop to reach a goal with little step-by-step guidance.
- API key
An API key is a secret code that lets a program call an AI service on your account, and the usage is billed to you.
B
- Batch API
A batch API takes many AI requests at once, processes them later and returns the results at a lower price than live calls.
- Benchmark
An AI benchmark is a fixed set of tasks with a scoring rule, used to compare models on the same test.
C
- Coding agent
A coding agent is an AI tool that reads a codebase, edits files and runs commands and tests in a loop to finish a programming task.
- Context window
The context window is the maximum amount of text, counted in tokens, that a model can read and write in one request.
E
- Embedding
An embedding is a list of numbers that represents the meaning of a text or image, so that similar items end up close together.
- EU AI Act
The EU AI Act (Regulation 2024/1689) is the European Union's law that regulates AI systems by risk level, with duties that start in stages.
F
- Fine-tuning
Fine-tuning trains an existing model further on your own examples, so it follows a style, format or task more reliably.
H
- Hallucination
A hallucination is a fluent, confident AI answer that is false, invented or not backed by the sources it was given.
I
- Inference
Inference is the act of running a trained AI model to produce an output from your input, such as an answer to a prompt.
K
- Knowledge cutoff
The knowledge cutoff is the date after which a model's training data stops, so it knows nothing about later events unless you give it fresh information.
L
- Large language model (LLM)
A large language model (LLM) is an AI system trained on huge amounts of text to predict the next token, which lets it write, summarise and answer.
- Local model
A local model is an AI model that runs on your own computer or server instead of a provider's cloud, so prompts need not leave your device.
M
- MCP (Model Context Protocol)
MCP (Model Context Protocol) is an open standard for connecting AI applications to external data, tools and workflows through one shared interface.
- Multimodal
A multimodal AI model can take in, and sometimes produce, more than one kind of data, such as text, images, documents, audio or video.
O
- Open-weight model
An open-weight model is an AI model whose trained weights are published for download, so you can run and often adapt it yourself.
P
- Parameters
Parameters are the numbers, also called weights, that a model adjusts during training and then uses to answer; their count is a rough measure of size.
- Price per million tokens
Price per million tokens is the usual way AI APIs state their rates: what you pay for each million tokens sent in or generated.
- Prompt
A prompt is the input you give an AI model: your question or task, plus any instructions, examples and text it should use.
- Prompt caching
Prompt caching lets an AI provider reuse the already processed start of your prompt, so repeated parts are cheaper and faster.
- Prompt injection
Prompt injection is an attack where text a model reads, such as a web page or file, contains instructions that make it act against its user.
R
- RAG (retrieval-augmented generation)
RAG (retrieval-augmented generation) first looks up relevant passages in your documents, then gives them to the model so it answers from them.
- Rate limit
A rate limit caps how many requests or tokens you may send to an AI API in a set time; going over returns an error, usually HTTP 429.
- Reasoning model
A reasoning model spends extra tokens on hidden step-by-step thinking before it writes its answer, which helps on hard problems but costs more.
S
- System prompt
A system prompt is a block of standing instructions, set by an app or developer, that shapes how a model behaves across a whole conversation.
T
- Temperature
Temperature is a setting that controls how random a model's word choices are: low gives steady answers, high gives more varied ones.
- Token
A token is a small chunk of text, often a word piece, that an AI model reads, writes and bills.
- Tool use (function calling)
Tool use, also called function calling, lets a model request an outside action, such as a search or a calculation, which an app runs and feeds back.
U
- Usage limit
A usage limit caps how much you can use an AI chat plan in a set period, such as a session or week, before you must wait or pay more.
Read next
- Token
A token is a small chunk of text, often a word piece, that an AI model reads, writes and bills.
- Context window
The context window is the maximum amount of text, counted in tokens, that a model can read and write in one request.
- Prompt
A prompt is the input you give an AI model: your question or task, plus any instructions, examples and text it should use.
- Hallucination
A hallucination is a fluent, confident AI answer that is false, invented or not backed by the sources it was given.