Token
The unit of text an AI model processes. Approximately 3/4 of a word.
Why it matters
Tokens are the currency of AI — you pay per token. Understanding them helps estimate costs.
In practice
Every LLM call logs tokens_in and tokens_out. Claude Opus 4.6 processes 1 million tokens in a single context.
Related terms
Token Budget
A limit on how many tokens an agent may use for a task.
Cost Tracking
Monitoring every AI call: model used, tokens in/out, cost, cache status. Essential for profitable operations.
Context Window
The AI model's "working memory" — how much information it can process at once (e.g., 1 million tokens for Claude Opus 4.6).
Inference
The process of an AI model generating a response or prediction from input data.