Large Language Model (LLM)

What is a Large Language Model?

A large language model (LLM) is a type of AI system trained on enormous quantities of text — books, websites, code, articles, and more — with one core job: predict what word (or token) comes next. Do that prediction billions of times across billions of examples, and something remarkable emerges: the model develops a working understanding of language, facts, reasoning, and even code.

When you send a message to ChatGPT, Claude, or Gemini, you are talking to an LLM.

How LLMs Learn

Training an LLM involves feeding it a huge corpus of text and asking it to repeatedly guess the next word. Each wrong guess slightly adjusts the model's internal parameters. After enough adjustments across enough text, the model gets very good at predicting plausible, coherent continuations — which turns out to be surprisingly close to understanding language.

The "large" in LLM refers to the sheer scale: billions or even trillions of numerical parameters, trained on terabytes of text.

Tokens, Not Words

LLMs do not read words — they read tokens, which are roughly word fragments. "Unbelievable" might be three tokens. This matters because model limits (and pricing) are measured in tokens, not words.

Well-Known LLMs

• GPT-4 / GPT-4o — OpenAI
• Claude 3.5 / Claude 4 — Anthropic
• Gemini 1.5 / 2.0 — Google
• Llama 3 — Meta (open weights)

LLMs are the foundation that powers chatbots, coding assistants, AI agents, and almost every AI product you interact with today.