Lesson 1143 of 1570
AI and tokens vs words: why your prompt costs what it costs
Learn what a token actually is so you can predict cost and context limits.
Lesson map
What this lesson covers
Learning path
The main moves in order
- 1The big idea
- 2What's a Token? Why ChatGPT Charges by Word-Pieces, Not Words
- 3The big idea
- 4What 'Tokens' and 'Context Window' Actually Mean
Concept cluster
Terms to connect while reading
Section 1
The big idea
AI doesn't see words — it sees tokens, which are usually 3-4 characters each. Knowing this explains why some prompts feel slow, hit limits, or cost more than expected.
How to use it
- Ask AI to count tokens in a sample paragraph
- Ask AI to explain why 'antidisestablishmentarianism' is many tokens
- Ask AI to translate 1000 words to a token estimate
- Ask AI to compare token efficiency across English vs Japanese
Try it
Paste a paragraph into a token counter (or ask AI). Predict the count first, then check how close you got.
Key terms in this lesson
Section 2
What's a Token? Why ChatGPT Charges by Word-Pieces, Not Words
Section 3
The big idea
Every AI model — ChatGPT, Claude, Gemini — slices your text into tokens before processing. One word is sometimes one token, sometimes three. The token is the unit of cost AND memory.
Some examples
- 'Hello' = 1 token, 'Antidisestablishmentarianism' = 6 tokens
- Emojis = 2-4 tokens each
- Code is more tokens than prose for the same characters
- GPT-4o input = $2.50 per million tokens; one essay ≈ 800 tokens
Try it!
Visit OpenAI's tokenizer (platform.openai.com/tokenizer). Paste your last essay. See how many tokens it actually was — and what that costs.
Section 4
What 'Tokens' and 'Context Window' Actually Mean
Section 5
The big idea
AI doesn't read your prompt as words — it reads it as 'tokens,' which are chunks roughly 3-4 characters long. Every model has a 'context window' (the max tokens it can hold in mind at once). When you exceed it, the oldest stuff gets dropped or summarized — which is why long chats get worse and earlier instructions get forgotten.
Some examples
- GPT-4o: 128,000 token context window — about 300 pages of text.
- Claude Sonnet 4.5: 200,000-1,000,000 tokens depending on tier — up to 2,500 pages.
- Gemini 1.5 Pro: 2,000,000 tokens — about 1.5 hours of video or 5,000 pages of text.
- When you upload a 50-page PDF and ask 5 follow-ups, you're using ~30k tokens before the model even reads your question.
Try it!
Visit platform.openai.com/tokenizer (free tool). Paste any paragraph and see how it breaks into tokens. Notice that 'unbelievable' is 4 tokens, but 'I' is 1. That's why prompt cost varies so much by language and topic.
Key terms in this lesson
End-of-lesson quiz
Check what stuck
15 questions · Score saves to your progress.
Tutor
Curious about “AI and tokens vs words: why your prompt costs what it costs”?
Ask anything about this lesson. I’ll answer using just what you’re reading — short, friendly, grounded.
Progress saved locally in this browser. Sign in to sync across devices.
Related lessons
Keep going
Builders · 40 min
What a Token Actually Is (And Why It Matters for Your Prompts)
AI doesn't read words — it reads tokens. Knowing the difference makes you a better prompter.
Builders · 40 min
Why AI 'Forgets' Halfway Through a Long Chat
AI has a memory limit called the context window. Hitting it explains a LOT of weird behavior.
Explorers · 40 min
Why AI Forgets the Start of a Long Chat
AI has a memory limit for how much of a chat it can remember at once.
