AI glossary
Start with the word you saw in a guide, a tool or a plan. Each entry gives you a short meaning. Five terms have a page with an example, so you can see how the word fits a task at work.
Facts checked on November 18, 2026.
AI terms from A to Z
- add-on
- An extra part of a tool that can do a task or link to another tool.
- AI
- Technology that lets computers do tasks like reading text or spotting patterns.
- AI chatbot
- An AI tool that answers you in a chat.
- arXiv
- A site where researchers share their papers.
- branch
- A separate line of work in the code. Teams can review changes on a branch before they add them to the main copy.
- ChatGPT Work
- A tool inside ChatGPT for carrying out work that can take several steps.
- cloud agent
- An agent that runs on remote computers you reach over the internet.
- cloud AI
- AI that runs a model on computers you reach over the internet.
- Codex
- An OpenAI tool that can work on code.
- coding agent
- An AI tool that works on code for you. You give it a goal, and it makes the changes.
- context window
- A context window is the space for what a model sees and writes.
- developer preview
- An early version that the maker has released for developers to try.
- harness
- The program around a model. It lets the model open files, run commands and use tools.
- IDE
- The program where people write code, also called a code editor. VS Code and the JetBrains editors are well-known ones.
- inference
- Inference is when a trained AI model uses new data to give a result.
- license
- The legal rules for who can use, copy or change the code. MIT and Apache-2.0 are two common open source licenses. Proprietary code can have tighter rules about who can use or change it.
- local AI
- AI that runs a model on a computer you control.
- local model
- An AI model that runs on your own computer instead of on a company's servers.
- model
- The part of an AI tool that uses what it learned to give an answer. Claude and GPT are model families.
- model training
- The process that teaches a model to find patterns in examples.
- open source
- The code is public and its license lets anyone use, change and share it under stated rules.
- open-weight model
- An open-weight model is an AI model whose maker shares the numbers it learned.
- plan limit
- A rule about how much you can use a tool on a given plan.
- prediction
- A guess based on what the model learned.
- provider
- A company that offers AI models you can use, such as OpenAI, Anthropic or Google.
- quantization
- Quantization stores model numbers with less detail so they take up less room.
- research preview
- An early version a maker shares so people can try it and report how it works.
- rollout
- When a company gives people access to a change in stages.
- SDK
- A kit that lets developers build their own tools on top of another tool.
- server
- A computer that does work or holds files for other computers.
- terminal
- A text window where you type commands to your computer. It is also called the command line.
- tokens
- A token is a piece of text an AI model reads or writes.
- Transformer
- A way to build a model that weighs how the parts of what you give it relate to each other.
- translation
- Changing text from one language to another.
- weights
- The numbers a model learns during training.
- working memory
- The space a computer uses to hold what it is working on right now.
- workspace
- A shared place in an app where a team can work and manage access.
Keep reading
- Claude vs ChatGPT
Compare plans, tools and how each company handles your chats.
- Local AI vs cloud AI
Find out where AI runs and what your team needs to check.
- History of AI
Trace three dates from an early AI workshop to ChatGPT.
- How AI coding agents work
Learn what lets an AI model work on files and run tasks.