# Token

> A token is the unit of text a language model reads and writes, usually a word, part of a word or a punctuation mark. Context windows are measured in tokens, and providers bill API usage in tokens, typically with separate rates for input and output. A common rule of thumb is that 1,000 tokens is roughly 750 English words, though the ratio varies by model and language.

## Why it matters for your business

Tokens are the meter on AI running costs. A workflow's monthly model bill is roughly the number of runs multiplied by the input and output tokens per run, multiplied by the model's per-token rates. Long system prompts, large retrieved documents and verbose outputs add up quickly at volume. Knowing token counts lets you estimate costs before launch, compare models fairly and spot waste, such as resending a full document on every step of an agent. Output tokens are usually priced higher than input tokens, so concise outputs save money.

## Example

A support team estimates the cost of AI reply drafting. Each ticket sends about 2,000 input tokens (instructions, the customer thread and two help articles) and gets back about 300 output tokens. At 15,000 tickets a month, that is about 30 million input tokens and 4.5 million output tokens, which the team multiplies by the chosen model's current rates.

Related terms: https://www.lumeor.studio/glossary/context-window/, https://www.lumeor.studio/glossary/inference/, https://www.lumeor.studio/glossary/prompt-caching/, https://www.lumeor.studio/glossary/model-routing/, https://www.lumeor.studio/glossary/ai-tco-total-cost-of-ownership/

---

Source: https://www.lumeor.studio/glossary/token/