# Tokens, context windows, and cost

> Learn why usage is measured in tokens rather than users.

Canonical: https://ainextia.com/learn-ai/tokens-and-cost/
Publisher: https://ainextia.com/

Level: Beginner
Track: Foundations
Reviewed: 2026-11-05

## Part 1
Text models process tokenized inputs and generate tokenized outputs. Tokens are pieces of text, not words. Different providers can count them differently, which matters when interpreting usage rankings.

## Part 2
A context window limits how much information a model can use in one request; a bigger advertised context does not automatically imply better recall. API billing may charge input, output, caching, and tool use differently.

## Part 3
Estimate how a long report affects your workload, then compare model provider pricing pages and check whether output length drives costs.

## Knowledge check
Why is 'most tokens processed' not equivalent to 'most people using a model'?

Reference: https://openrouter.ai/rankings
