Paste any text or prompt and see exactly how many tokens it uses — with the same tokenizer ChatGPT uses — plus whether it fits popular context windows and what it costs through the API. Free, no signup, and your text never leaves your browser.

⌛ Loading the exact ChatGPT tokenizer… counts below are calibrated estimates until it's ready.
0
tokens
estimated
0
words
0
characters
Does it fit the context window?
What would it cost through the API?
ModelInput $/1MOutput $/1MAs inputAs output
1 token ≈ 5 characters of English 100 words ≈ 115 tokens A 128K context window ≈ a full novel

Token counts use the o200k encoding — the one used by GPT-4o and newer OpenAI models. Claude and Gemini use their own tokenizers; their counts are typically within roughly 10–20% of these numbers. API prices verified July 2026 and editable above — always confirm on the provider's pricing page. Everything runs in your browser; nothing you paste is sent to a server.

Spending Tokens on Prompts That Don't Deliver?

A well-built prompt gets better answers in fewer tries. The ChatGPT Prompt Pack gives you 4,008 tested prompts organized by category — or build your own with our free AI Prompt Generator.

Get the Prompt Pack →

What Is a Token, and Why Should You Care?

Language models like ChatGPT don't read words — they read tokens: chunks of text that are sometimes a whole word, sometimes part of one, sometimes just punctuation. "Hello, how are you today?" is six words but seven tokens. Tokens matter for two practical reasons. First, context limits: every model has a maximum number of tokens it can hold in one conversation, and when you exceed it, the model starts forgetting the beginning. Second, cost: API pricing is per token, for both what you send and what the model writes back — so knowing your token count is knowing your bill. Click "Show token breakdown" above to see exactly how the model slices your text; it's the fastest way to build an intuition for it.

Frequently Asked Questions

Is this token counter accurate?

Yes — once loaded, it uses the real o200k encoding, the same tokenizer GPT-4o and newer OpenAI models use, so the count is exact, not an approximation. While the tokenizer library loads (it's about 2 MB), the tool shows a calibrated estimate and tells you which mode you're seeing.

Do Claude and Gemini count tokens the same way?

No — each provider has its own tokenizer, so the exact numbers differ slightly. In practice the counts are usually within 10–20% of each other for English text, so this tool gives you a reliable ballpark for any model.

Is my text stored or sent anywhere?

No. The tokenizer runs entirely in your browser — your text never leaves your device, which also makes this safe for confidential documents.