FREE TOOL · DEVELOPER
Estimate tokens before you send a prompt
AI models read and bill in tokens — chunks of roughly four characters of English. Paste a prompt to estimate its tokens and context use, then shrink a bloated prompt in the challenge without losing meaning.
You will practise
- What tokens are
- Estimating context-window use
- Writing compact prompts
⚙Use the tool
🔒 Runs entirely in your browser. Nothing is uploaded.
✓Practice challenge
Rewrite this prompt in 60 tokens or fewer, keeping “customer”, “refund” and “48”:
I would really like you to please help me write a reply to a customer who has written to us because they are not happy and they want a refund for their order, and I want you to tell them in a very polite way that we will process the refund within 48 hours if they send us the receipt.
✓ Challenge recorded on this device.
Questions people ask
How accurate is this counter?
It is an estimate (about 4 characters per token for English). Exact counts depend on each model’s tokenizer; other languages often use more tokens.
Why do tokens matter?
They decide cost, speed and whether your text fits the model’s context window.
Keep building
DeveloperAI API Cost CalculatorCalculate what an AI feature will cost per monthAI skillsPrompt GeneratorBuild a prompt with role, task, context and checksDeveloperJSON Formatter & ValidatorFormat, validate and fix JSONACADEMY COURSEAI, without the guessworkGo deeper: a full project course with fixtures, a capstone and a handbook.