What is Token?
Token: The unit of text an LLM reads and writes, roughly a word fragment. Model pricing, speed, and context limits are all measured in tokens.
Language models do not see words or characters. They see tokens, which are chunks of text from a fixed vocabulary. In English, a token averages around three quarters of a word, or about four characters, though it varies by language and by model. Code, numbers, and non-English text often use more tokens per word.
Tokens matter in three ways. Cost: providers bill per million input tokens and per million output tokens, and output is usually priced several times higher than input. Limits: the context window is a token budget for everything the model sees at once. Speed: output is generated token by token, so longer responses take longer, which is critical for voice and chat.
Why it matters when you build or buy AI
Token usage is the main variable cost in most AI products, and it is easy to waste. Stuffing whole documents into every request, resending long histories, and asking for verbose answers can multiply your bill without improving results. Before you sign off on an AI build, ask for the expected tokens per task and the resulting cost per task at your volume.
Example
A 2,000-word contract is roughly 2,700 tokens. Summarizing it in 200 words costs that input plus about 270 output tokens per request.
Related terms
More in AI & LLMs
Book 20 minutes, or send a brief
A former CTO replies. You leave with a written plan and a fixed estimate.
Prefer LinkedIn? Connect with Harish.