Start a project

SUPERR_dictionary

The terms people use about AI, defined in plain words, sourced and dated.

44 entries · 5 categories · A–Z

5 Word of the day Thursday · November 2026 New word at
00:00 New York

Model Context Protocol

Mod·el Con·text Pro·to·col /ˈmɑ.dəl ˈkɑnˌtɛkst ˈproʊ.təˌkɔl/ noun Practice

The Model Context Protocol (MCP) is an open standard, launched by Anthropic in 2024, that lets AI applications connect to outside tools and data.

An AI assistant connects through one standard socket to three servers labeled calendar, database and code. A dashed tangle of custom cables beside it is crossed out. — Inkling · One standard plug in place of a custom cable for each toolFig. M1

Read the entry →

Most looked up

Editors’ list until there are 30 days of views

Recently added

Newest first

Browse by category

Colour is category, on every page

All entries

44 entries · A–Z

A

4 entries

B

1 entry

benchmark Practice In AI, a benchmark is a standard test, the same tasks scored the same way for every model, used to compare models and track progress over time.

C

4 entries

D

2 entries

E

2 entries

F

2 entries

G

3 entries

H

2 entries

I

1 entry

inference Foundations In AI, inference is the stage in which a trained model is put to use, turning new inputs into outputs such as answers or predictions.

J

1 entry

jailbreak Safety In AI, a jailbreak is a prompt that gets a model to ignore its safety training and produce content it was built to refuse.

K

1 entry

knowledge cutoff Foundations In AI, a knowledge cutoff is the date a model’s training data ends; after it, the model knows little unless newer facts are supplied.

L

1 entry

large language modelLLM Foundations In AI, a large language model (LLM) is a neural network trained on huge amounts of text to predict the next word, the engine behind chatbots like ChatGPT.

M

2 entries

O

2 entries

P

2 entries

Q

1 entry

quantization Foundations In AI, quantization is storing a model’s weights in fewer bits, often 8 or 4 instead of 16, so the model uses less memory and runs faster.

R

4 entries

S

2 entries

T

4 entries

V

2 entries

W

1 entry

weights Foundations In AI, weights are the numbers a model adjusts during training, which set how it turns inputs into outputs and together make up the trained model.