The length of AI questions and answers is counted in tokens. The number of tokens determines the cost and the model’s context limit.
What it means in practice
Tokens are small pieces of text – parts of words, whole words or characters – into which an AI model splits text before processing it. The length of a question, context and answer is measured in tokens; in Polish one word averages 2–3 tokens.
Why it matters for your business
The number of tokens determines the cost of using AI via an API and how much information the model can “remember” at once (the context window). That’s why AI modules carry a small monthly usage fee.
Examples and good practice
- Short, specific prompts are cheaper and often more effective.
- Long documents can exceed the model’s context limit.
- Polish uses more tokens than English.
- Cost depends on the model – stronger models cost more per token.
How Novi handles it
At Novi token costs for AI modules (assistant, review replies, content) are included in a fixed monthly fee of PLN 25–75. You don’t have to count anything or fear surprises on your invoice.
