Token Counter
Paste text, get the token count — with a real GPT tokenizer running in your browser — plus what sending that text costs across the major model APIs.
1 · Your text
2 · What this text costs as input, per model
| Model | Input rate /1M | Cost to send this text once | ×1,000 sends |
|---|
Cost uses each model's published input rate against the token count above (GPT models use the exact o200k count; others use the labelled estimate). Output tokens are billed separately — plan full workloads with the API pricing calculator.
How counting works here
- GPT (exact). The page loads the open-source gpt-tokenizer library (o200k_base and cl100k_base encodings) from a CDN and tokenizes locally. If the CDN is unreachable, the page falls back to estimation and labels every figure as an estimate.
- Claude & Gemini (estimate). Anthropic and Google do not ship official in-browser tokenizers, so we approximate: Claude-family ≈ o200k count × 1.15, Gemini-family ≈ o200k count × 1.05 (different vocabularies typically tokenize English prose within ±15% of each other). These are planning figures, not billing figures.
- Rule of thumb. For English prose, 1 token ≈ 4 characters ≈ 0.75 words. Code, CJK text and unusual formatting tokenize less predictably.
Frequently asked questions
How many tokens is my text?
Roughly 4 characters or three-quarters of a word per token for English prose — but paste it above for a real count. GPT figures come from OpenAI's actual tokenizers running in your browser.
Is a Claude token the same as a GPT token?
No — every model family has its own tokenizer, so the same text yields different counts. That's why the Claude and Gemini numbers here are labelled estimates.
Is my text uploaded anywhere?
No. Tokenization runs entirely client-side in JavaScript; nothing is sent to a server.
Why does the cost table matter?
Because the same prompt costs 70x more on GPT-5.5 Pro than on DeepSeek V4-Flash. Token count × input rate is the first number to know before wiring a prompt into a loop — then run monthly volumes through the pricing calculator.