What is an LLM?
An LLM (large language model) is an AI model trained on huge amounts of text to predict the next piece of text. That one skill, at enormous scale, lets it answer questions, write, summarize, translate and code. ChatGPT, Claude, Gemini and Llama are all built on LLMs.
An LLM is autocomplete on steroids. Your phone guesses your next word; an LLM guesses the next word so well, thousands of times in a row, that it can write essays and code.
- You type a prompt"Why is the sky blue?"
- Chopped into tokensSmall word pieces the model can count
- Guess the next tokenPicks a likely next piece from everything it learned
- Add it, repeatOver and over, one token at a time
- Your answer appearsThat's why replies stream in word by word
It never plans the whole answer — it just keeps predicting the next piece.
How an LLM works, simply
Text is split into small chunks called tokens (roughly word pieces). During training, the model reads billions of sentences and keeps adjusting its internal numbers — its parameters, or "weights" — to get better at guessing the next token. After training, you give it a prompt and it generates a reply one token at a time.
Almost every modern LLM uses the transformer design introduced by Google researchers in the 2017 paper "Attention Is All You Need". The "large" refers to the size: billions of parameters and a training set covering a large share of the public internet, books and code.
What LLMs are good and bad at
LLMs are very good at language tasks: drafting, rewriting, explaining, summarizing and writing code. Their big weakness is that they generate plausible text, not verified facts, so they can state wrong things confidently — usually called "hallucination". They also only know what was in their training data unless they are connected to search or other tools.
LLM vs chatbot vs agent
The LLM is the engine. A chatbot like ChatGPT is a product wrapped around it. An AI agent is an LLM that is given tools — a browser, a code editor, your files — and allowed to take actions on its own, which brings new risks such as prompt injection.
▸ THE HYPE VS WHAT WE FOUND
Real claims involving this term, each checked in a PixProof with linked sources.
▸ PIXPROOFS THAT USE THIS TERM (4)
▸ COMPANIES BEHIND IT
HYPE OR REAL?
"The iPhone 17 Pro can genuinely run large AI models locally, on the phone itself."
The answer: Real. Independent testing backs Apple's claim — it runs large local AI models well.
See the full iPhone 17 Pro PixProof → Play 12 more in Buy or Bye? →▸ QUICK QUESTIONS
- What does LLM stand for?
- Large language model.
- Is ChatGPT an LLM?
- ChatGPT is a chatbot product; the LLM underneath it is OpenAI's GPT family of models.
- Why do LLMs make things up?
- They are trained to produce likely-sounding text, not to check facts, so when they lack the information they can still produce a fluent but wrong answer.
▸ SOURCES
- Wikipedia — Large language model ↗
- Vaswani et al. — Attention Is All You Need (2017) ↗
- Wikipedia — Hallucination (artificial intelligence) ↗
Spot something wrong? Tell us — corrections are logged, see methodology.