Growthtrait
All terms
AI Token (LLM Token)

Token

A token is the basic unit of text a large language model reads and writes. Before a model processes a prompt, it splits the text into tokens, which can be whole words, parts of words, punctuation or spaces. Tokens are how models measure input length, output length, context limits and price.

People read words. AI models read tokens. A tokenizer breaks text into pieces the model has learned, so a common word may be a single token while a rare or long word is split into several. Everything about using an LLM is counted in tokens: how much you can send, how much it can write back, and how much you pay.

How Many Words Is a Token?

For English, a common rule of thumb is that one token is about four characters, or roughly three quarters of a word, so 1,000 tokens is around 750 words. The exact count depends on the model's tokenizer. Languages with accents or non Latin scripts, including Vietnamese, often use more tokens for the same meaning, which makes the same content more expensive to process.

Why Do Tokens Matter?

  • Price: AI APIs charge per million tokens, usually with separate prices for input and output tokens. Output tokens tend to cost several times more than input.
  • Limits: every model has a maximum context window, measured in tokens, covering everything it can consider at once.
  • Speed: the more tokens a model has to read and write, the longer a response takes.

Input, Output and Cached Tokens

Input tokens are everything you send: instructions, documents, chat history and tool definitions. Output tokens are what the model writes back, including hidden reasoning tokens on some models. Many providers also offer cached tokens, a much cheaper rate for input that repeats across requests, such as a long system prompt. In our GPT-6 Astra vs Claude Fable 5.1 comparison, both models list around $10 per million input tokens and $50 per million output tokens, and the real cost difference came from cached token pricing.

How to Use Fewer Tokens

  • Send only the context the task needs, the core idea of context engineering.
  • Keep stable instructions at the start of the prompt so they can be cached.
  • Ask for the output format and length you actually need.
  • Route simple tasks to smaller, cheaper models with an AI router.

Frequently asked questions

What is a token in AI?+

A token is a small piece of text, such as a word, part of a word or a punctuation mark, that an AI model uses as its basic unit for reading, writing, limits and pricing.

How many words are in 1,000 tokens?+

In English, 1,000 tokens is roughly 750 words. The exact number depends on the model's tokenizer and the language, and languages like Vietnamese usually need more tokens per word.

Why are output tokens more expensive than input tokens?+

Writing output takes more computation than reading input, because the model generates each output token one after another. Providers price that extra work into output tokens.

Related terms

Context Window

LLM Context Window

A context window is the maximum amount of information, measured in tokens, that a large language model can take into account in a single request. It includes the instructions, the conversation so far, any documents or tool results, and the answer the model writes. Anything outside the window, the model simply cannot see.

LLM

Large Language Model

An LLM (Large Language Model) is an AI model trained on vast amounts of text to understand and generate human language. It works by predicting the most likely next piece of text given everything so far, which lets it answer questions, write, summarize, and hold a conversation.

Context Engineering

Context Engineering

Context engineering is the practice of curating and maintaining the right set of information an AI model sees at each step of a task, rather than only phrasing a single prompt. It covers the whole information environment around the model, including instructions, retrieved documents, memory, and tools.