What is Tokens (AI)?
The text chunks an AI reads and counts in — roughly words, but split up. And they are usually what you pay for.
A token is a small piece of text that an AI works with. When you send text to an AI, it is split into tokens — sometimes a whole word, sometimes just part of a word or a punctuation mark. An AI does not read letters or sentences; it processes these chunks, one at a time.
A simple rule of thumb: one token is roughly three quarters of a word. The word "hi" might be one token, while a long word like "organisation" might become several. You do not need to count them yourself, but it is worth knowing they exist, because they are the unit the AI measures everything in.
Why does it matter? Because tokens are often what you pay for when you use AI services. The price is set per number of tokens in and out — a bit like paying by the metre instead of a flat fee.
Why are tokens important for your business?
If you use AI at scale, tokens directly affect what it costs. Long prompts and long answers use more tokens and therefore more money. Understanding that helps you keep costs down without cutting quality.
There is also a practical limit: each model can only handle a certain number of tokens at once. That determines how much text you can feed in at one time — for example, how long a document an AI can summarise in a single go.
Tokens in practice
Say you want an AI to summarise your customers' long support emails every day. The longer the email, the more tokens, the higher the cost. A good solution might be to shorten the text sent in, or only send the relevant parts — so you get the same value but pay less.
Common questions about Tokens (AI)
What is a token in AI?
A token is a small piece of text that an AI reads and writes in, roughly a word or part of a word. The AI processes text one token at a time. Tokens are also often what you pay for when using AI services.
How many words is a token?
As a rule of thumb, one token is roughly three quarters of a word. Short words often become one token, while long words may be split into several. It varies by language and word.
Why do tokens matter for cost?
AI services usually charge per number of tokens in and out. Longer texts use more tokens and therefore cost more. That is why it pays to keep prompts and answers reasonably short.
Related terms
AI Agent
AI that can carry out tasks for you — not just answer questions. Books meetings, sends emails, makes decisions.
AI Agent vs Chatbot
The chatbot answers questions. The AI agent goes and fixes it. One talks; the other rolls up its sleeves.
Context Window
How much text an AI can hold in memory at once. A larger window means more context — some models now handle millions of tokens.
Copilot
An AI assistant that works alongside you and suggests next steps. You are still the captain — it is the co-pilot, not autopilot.
Embedding
Turning text into numbers so AI can compare meaning. The foundation for smart search and chatbots.
Fine-Tuning
Training an AI model on your specific data. Expensive but powerful — only when you really need it.