Token calculator: count tokens and see what a text costs
Last verified: — every figure checked against the official page linked next to it.
AI models don’t read words, they read tokens: pieces of words, about five characters each with today’s tokenizers. Prices, context windows and usage limits are all measured in tokens. Paste a prompt, an email or a document to count its tokens and see what it costs in each model. Your text never leaves your browser.
Loading the tokenizer… showing an estimate (characters ÷ 4) for now.
What this text costs to send, by model
Your text as input plus an answer of the length chosen above, at standard API prices. Prices from our AI pricing page. →
| Model | One request | 1,000 requests |
|---|---|---|
| GPT-6 Luna · OpenAI | $0.00031 | $0.3059 |
| Claude Haiku 5.5 · Anthropic | $0.00031 | $0.3059 |
| Mistral Small 4 · Mistral | $0.00037 | $0.3689 |
| DeepSeek V4.1 Flash · DeepSeek | $0.00074 | $0.7377 |
| Mistral Large 4 · Mistral | $0.00129 | $1.29 |
| Gemini 3.5 Flash-Lite · Google | $0.00152 | $1.52 |
| Grok 4.3 · xAI | $0.00157 | $1.57 |
| Gemini 3.8 Flash · Google | $0.00229 | $2.29 |
| DeepSeek V4 Pro · DeepSeek | $0.00245 | $2.45 |
| Grok 4.7 · xAI | $0.00372 | $3.72 |
| Mistral Medium 3.5 · Mistral | $0.00459 | $4.59 |
| Gemini 4 Argon · Google | $0.00612 | $6.12 |
| GPT-6.1 Sol · OpenAI | $0.00612 | $6.12 |
| Claude Sonnet 5.5 · Anthropic | $0.00612 | $6.12 |
| Gemini 3.1 Pro (preview) · Google | $0.00732 | $7.32 |
| Claude Opus 5.5 · Anthropic | $0.0122 | $12.24 |
| GPT-6 Astra · OpenAI | $0.0306 | $30.59 |
| Claude Fable 5.1 · Anthropic | $0.0306 | $30.59 |
Does it fit in the context window?
How much of each chat plan’s context window this text fills. Every reply in the same chat adds to it. →
| Plan | Window | Used by this text |
|---|---|---|
| ChatGPT Free | 27,000 | 0.2% |
| Gemini (no plan / sin plan) | 32,000 | 0.2% |
| ChatGPT Go / Plus (Instant) | 54,000 | 0.1% |
| Gemini · Google AI Plus | 128,000 | <0.1% |
| ChatGPT Go / Plus (reasoning) | 256,000 | <0.1% |
| Claude · Gemini AI Pro / Ultra | 1,000,000 | <0.1% |
What is a token?
A token is the unit a language model reads and writes. Common short words are a single token; longer or rarer words are split into pieces, and spaces and punctuation count too. With OpenAI’s current tokenizer, one token is about five characters of ordinary English text, a little under one word.
Tokens matter because everything is measured in them. API prices are per million tokens, separately for what you send (input) and what the model writes (output). The context window, how much a chat can hold, is a number of tokens. And when a chat app talks about usage limits, longer messages and longer conversations use up more of it.
Spanish uses about 15% more tokens than English
We tokenized the same four prompts written in English and in Spanish. The English versions came to 176 tokens and the Spanish ones to 203: 15% more for the same content. Per word, English averaged 1.15 tokens and Spanish 1.23.
The practical effect is small but real: the same request costs a little more, and fills the context window a little faster, in Spanish than in English. Clear, specific prompts save far more than switching language would.
Rules of thumb
- 1 token ≈ 5 characters of English text (5.2 in our test) and 4.9 in Spanish.
- 1,000 tokens ≈ 870 English words or 810 Spanish words, about two pages.
- A one-page email or prompt is usually 300–600 tokens.
- In a chat, each new message resends the whole conversation, so input tokens grow with every turn.
Frequently asked questions
How many tokens is 1,000 words?
In our test, about 1,150 tokens in English and 1,230 in Spanish. Spanish also needs more words to say the same thing, so the same text ends up around 15% more tokens. Older tokenizers produced more tokens per word. Paste your own text above for an exact count.
Do ChatGPT, Claude and Gemini count tokens the same way?
No. Each provider has its own tokenizer, so the same text gives slightly different counts. This calculator uses o200k_base, the tokenizer of OpenAI’s current models, which is a good estimate for the others.
Is my text sent anywhere?
No. The counting runs entirely in your browser. Nothing you paste is uploaded or stored.
Why does a long chat cost more than a short one?
Because each new message is sent together with the whole conversation so far. A chat with ten exchanges sends all ten again on the eleventh message, which is why retries and corrections add up quickly.