Term: Token
~6 min read
Estimated time: ~6 min read — for the in-app brief plus opening the primary source.
What this is
A token is the small unit of text a model reads and writes — often a short word or part of a word. Limits and most API bills are counted in tokens.
Everyday example
A 50-page agreement is thousands of tokens. Paste it into ChatGPT, Claude, or Grok and you may hit a limit, pay more, or lose early clauses. The model is counting tokens — not pages.
A token is the small unit of text the model counts — not pages, not words exactly.
- Roughly: a short word or part of a long word is a token.
- Your input and the model’s reply both spend the same token budget.
- Long pastes hit limits, drop early clauses, or raise cost.
- Vendors price many APIs per million tokens.
Next action: On the next usage quote, ask for expected tokens per task × monthly volume — not seats only.
What changes in how you lead
How decision rights, process, and ownership should change.
- Legal and HR workflows that paste entire files are cost and quality designs.
- Purchasing compares unit economics in tokens, then translates to process cost.
Compare related ideas
Token vs Context window
Token is the unit counted. The context window is how many tokens fit in one sitting — your input plus the model’s reply.
Open Context windowToken vs Token pricing
Tokens measure volume. Token pricing turns that volume into a bill. A cheap model on a huge paste can still cost more than a dear model on a tight brief.
Open Token pricingDeep dive
Models do not see pages. They see a budget of tokens for what you send in and what they write back.
Legal: long agreements need section-by-section work or retrieval — not one giant paste.
HR: screening hundreds of résumés is a token-volume and cost problem, not only a quality problem.
Purchasing and Finance: compare tools on expected tokens per task × monthly volume, not seat price alone.
Marketing: send the relevant brand sections, not the entire book, so the model spends tokens on the right material.