The Overlord's Lexicon
What is GPT?
Also known as: Generative Pre-trained Transformer
GPT stands for Generative Pre-trained Transformer, a family of large language models developed by OpenAI that power ChatGPT. The term is also used loosely for any similar text-generating AI model.
GPT is the most famous three-letter acronym in AI. It names OpenAI’s series of language models — the engine inside ChatGPT — and has become shorthand for chatbot-style AI in general, the way “Hoover” became a verb for vacuuming.
What the letters mean
- Generative: it produces new text (and, in recent models, images and audio) rather than just classifying it. See generative AI.
- Pre-trained: it first learns general language patterns from a massive corpus of text before being tuned for specific behaviour, like following instructions and chatting safely.
- Transformer: it’s built on the transformer architecture introduced by Google researchers in 2017.
OpenAI released the first GPT model in 2018. Successive versions grew dramatically in size and capability, and the launch of ChatGPT in November 2022 turned the technology into a mainstream product almost overnight.
An analogy for the meatbags
GPT is to ChatGPT what an engine is to a car. OpenAI keeps building new engines (GPT models) and fitting them into the car you actually drive (ChatGPT). Other manufacturers build their own engines — Anthropic’s Claude, Google’s Gemini, Meta’s Llama — which is why “GPT” isn’t technically the right word for all of them, however often humans use it that way.
Why it matters for the Chosen
Understanding the distinction saves you from confused buying decisions and bad advice:
- Models vs products. The same model can be accessed through ChatGPT, through OpenAI’s API, or inside third-party tools. Many “AI writing apps” are GPT (or Claude, or Gemini) behind a custom interface — an AI wrapper.
- Choosing tools. GPT models are excellent general-purpose options, but for any given task — long documents, coding, writing style, cost — another model might be better. The Chosen test several.
- Custom GPTs. ChatGPT lets you build configured versions with your own instructions and files. They’re useful for packaging a repeatable workflow — an SEO brief writer, a client-onboarding helper — for yourself or your team.
Examples
ChatGPT runs on OpenAI’s GPT models. Microsoft’s Copilot products have used OpenAI models. Countless startups call GPT models through the API to power writing, support and coding tools. “GPT” also appears in unrelated product names, so check what model a tool actually uses before you trust the label.
Tools from the Arsenal
ChatGPT
AI WritingOpenAI's general-purpose assistant: writing, research, images, voice and agents in one chat box.
GitHub Copilot
AI CodingGitHub's AI pair programmer: completions, chat and coding agents across editors and GitHub itself.
Questions the Humans Keep Asking
What is the difference between GPT and ChatGPT?
GPT refers to OpenAI's underlying models. ChatGPT is OpenAI's chat product that runs on those models, adding a conversational interface, memory, tools and other features.
Is Claude a GPT?
Not in the brand sense — Claude is Anthropic's model family and GPT is OpenAI's. Technically, both are transformer-based large language models, which is why people sometimes use 'GPT' generically.
What are custom GPTs?
Custom GPTs are tailored versions of ChatGPT that users can configure with their own instructions, files and tools for a specific purpose, and optionally share with others.