What exactly is a 'token' in AI, and why should I care?

A token is a chunk of text that an AI model reads or generates, and you should care because most AI tools charge based on the number of tokens you use, not the number of words you see. It's the fundamental unit of cost and processing. Think of it like this: if words are the meal, tokens are the individual ingredients. The word 'apple' might be one token, but a more complex word like 'unbelievable' could be broken into three tokens: 'un', 'believ', and 'able'. A common rule of thumb is that 750 English words equal about 1,000 tokens, but this is a rough estimate. The real-world impact is on your wallet and your workflow. For example, if you're using a tool that charges $0.03 per 1,000 tokens, generating a 3,000-word blog post might cost you around $0.12. That sounds tiny, but it adds up fast with heavy use. The hidden cost comes from the conversation history. Every time you ask a follow-up question, the AI has to re-read the entire conversation, burning tokens on old messages. I've seen people burn through their monthly budget in a week because they didn't realize each back-and-forth was adding to the tally. A useful tip: start a new chat session for a new topic. It prevents the token meter from running on irrelevant history and keeps the AI focused. **Related**: How do I estimate the cost of an AI writing project? | What's the difference between input and output tokens?