// Learn

What is an LLM?

An LLM (Large Language Model) is an AI system trained on vast amounts of text that can generate, analyse, and reason about language.

The short version

When you talk to ChatGPT, Claude, or Gemini, you're talking to an LLM. These systems have read enormous amounts of text (books, websites, code, conversations) and learned patterns in how language works. They use those patterns to generate responses that are coherent, contextual, and often surprisingly useful.

They don't "understand" the way humans do. They predict the most likely next word, billions of times, very quickly. But the result is powerful enough to write code, explain concepts, translate languages, and reason through problems.

How it works

The key concepts:

  • Training: the model reads billions of words and adjusts millions (or billions) of internal parameters to predict what comes next. This takes weeks on thousands of specialised chips.
  • Tokens: LLMs don't read words; they read tokens (roughly ¾ of a word). "Understanding" becomes three tokens: "under", "stand", "ing".
  • Context window: how much text the model can consider at once. Early models handled ~4,000 tokens. Modern ones handle 100,000 to 1,000,000+.
  • Inference: when you send a prompt and get a response, that's inference. The model generates one token at a time, each informed by everything before it.
  • Temperature: a setting that controls randomness. Low temperature = more predictable responses. High temperature = more creative, more varied.

The "large" in LLM refers to the number of parameters, the internal dials the model adjusts during training. GPT-4 and Claude have hundreds of billions. More parameters generally means more capability, but also more cost to run.

Why it matters

LLMs are the foundation of the current AI wave. They power chatbots, coding assistants, search engines, and content tools. Understanding what they can and can't do (that they're pattern matchers, not thinkers) helps you use them more effectively and spot their limitations.

=++==+==++=

//Read more

aiWhat is MCP?Model Context Protocol is a universal standard that lets AI assistants use external tools.fundamentalsWhat is an API?APIs let software talk to other software. They're how everything on the internet connects.fundamentalsWhat is open source?Open source means the code is public. Anyone can read it, use it, and contribute.aiWhat are AI agents?AI agents are LLMs that can use tools, remember context, and take actions on your behalf.aiWhat is prompt engineering?Prompt engineering is the skill of writing instructions that get useful results from AI models.aiWhat are tokens?Tokens are the units AI models use to read and write text. They determine cost, speed, and limits.aiWhat is RAG?RAG gives AI models access to external knowledge so they can answer questions about your data.aiWhat is streaming in AI?Streaming is why AI responses appear word by word instead of all at once.aiWhat is vibe coding?Vibe coding is building software by describing what you want to AI, rather than writing code yourself.aiWhat is fine-tuning?Fine-tuning adapts a pre-trained AI model to perform better at a specific task or domain.aiWhat is hallucination in AI?Hallucination is when an AI generates something that sounds right but isn't true.aiWhat are embeddings?Embeddings turn text into numbers so machines can measure how similar two pieces of content are.