Glossary · 1 minute read
What Are Tokens in AI?
In AI, a token is a chunk of text—roughly a word or part of a word—that a language model reads and generates. Models process text as sequences of tokens, and both pricing and context limits are measured in tokens, not words. A rule of thumb is that one token is about three-quarters of an English word. Understanding tokens matters because they drive inference cost and constrain how much text a model can handle at once, shaping both budget and design.
Tokens are how language models read and price text. Here's what they are, why they drive cost and context limits, and what that means for building with LLMs.
What a token is
A token is a chunk of text—roughly a word or part of a word—that a language model reads and generates. Models process text as sequences of tokens, not words. A rule of thumb: one token ≈ ¾ of an English word (1,000 tokens ≈ 750 words).
Why tokens matter
| Impact | Why |
|---|---|
| Cost | Providers price per token |
| Context limit | Measured in tokens (context window) |
| Latency | More tokens, slower responses |
Tokens drive both cost and design constraints—central to generative AI cost and LLM application cost.
Tokens and cost
LLM providers charge for input and output tokens. Longer prompts and responses cost more—so concise prompts, retrieval to avoid stuffing context, and caching directly control inference cost.
Tokens and context limits
A model's context window is measured in tokens—the maximum it can consider at once. Long documents may exceed it, which is why RAG retrieves only the relevant chunks.
Why FISTA
FISTA Solutions builds LLM systems with token-aware design—managing cost and context deliberately—through AI enablement, backed by 150+ projects across 12+ countries.
Building cost-efficient LLM features? Talk to FISTA.
Share-ready article cover
Download the generated social format.
Clear answers
Questions raised by this field note.
Straightforward guidance for evaluating scope, fit, and the next step.
01What is a token in AI?
A chunk of text—roughly a word or part of a word—that a language model reads and generates. Models process text as sequences of tokens, and both cost and context limits are measured in tokens rather than words.
02How many words is a token?
Roughly, one token is about three-quarters of an English word, so 1,000 tokens is about 750 words. It varies by language and text, since common words may be one token and rare words several.
03Why do tokens matter for cost?
Because LLM providers price by tokens processed—both input and output. Longer prompts and responses cost more, so managing token usage through concise prompts, retrieval, and caching directly controls inference cost.
Continue exploring
Related capabilities
Start with the hard problem
Need the outcome owned, not merely analyzed?
Tell us where delivery is constrained. We’ll map the fastest credible path from intent to verified production.