# LLM (Large Language Model)

> An LLM is an AI model trained on massive amounts of text to predict the next word, which lets it answer questions, summarize, translate and write code.

## How it works

An LLM takes input text (the *prompt*), splits it into pieces called *tokens* and predicts, token by token, the most likely continuation. That simple mechanism, repeated billions of times during training, produces models that can follow instructions, reason step by step and write code.

## Where LLMs fail

- **Hallucination:** they produce plausible-sounding facts that don't exist.
- **Frozen knowledge:** they only know what was in the training data, up to a cutoff date.
- **Variable cost and latency:** longer answers cost more and take longer.

## In production

Running an LLM in production takes the same rigor as any critical system: latency and error metrics, cost per request, prompt-injection defenses and a plan for when the model provider goes down.