Skip to main content
ModelTerms

Comparison

Prompt vs Token

Prompt and Token are both common AI/LLM terms but cover different ideas. Here is a quick side-by-side.

When you would reach for Prompt

Prompt comes up when the question is fundamentally about prompting.

A system prompt + user message + a few-shot examples = a typical prompt.

When you would reach for Token

Always think in tokens, not characters, when planning prompts, budgets, and context windows.

"Hello, world!" tokenizes to roughly 4 GPT-4o tokens.

Frequently asked

What is the difference between Prompt and Token?

Prompt: A prompt is the text you send to an LLM to elicit a response. It typically includes a system message, optional examples, and the user's query. Token: A token is the basic unit an LLM reads and writes — usually a word piece (3-4 characters). LLMs are priced and sized by tokens, not words.

When should I use Prompt vs Token?

Prompt is the right concept when you are focused on prompting. Always think in tokens, not characters, when planning prompts, budgets, and context windows.

Are Prompt and Token the same thing?

No. Prompt is prompting; Token is inference. They are related but address different parts of the AI stack.