Claude Token Counter
Count tokens and API cost for Claude Opus 5.5, Sonnet 5.5 and Haiku 4.5, in any language.
The same text can use from 3% (Chinese) to 144% (Punjabi) more tokens than English — see how we measured 41 languages.
🔍 See how it splits into tokens
Each colour is one token (GPT-4o and later). Boxes with a dashed outline are a single character that takes several tokens.
Colours just split tokens. Same price each: more boxes = higher cost.
Official pricing & tokenizer references:
How do I count tokens for Claude?
Paste your text above with the Claude tab selected. The counter shows tokens, characters and the input and output cost on Claude Opus, Sonnet and Haiku. For an exact count from Anthropic, use the count_tokens endpoint of the Claude API.
Does Claude use the same tokenizer as GPT?
No. Claude has its own tokenizer. In our test it used 38% more tokens than GPT for the same English prompt, and the gap depends on the language. This counter applies the measured figure for each language to its estimate.
How much does Claude cost per token?
Per million input / output tokens: Claude Fable 5.2 $8 / $40; Claude Fable 5.1 (legacy) $10 / $50; Claude Opus 5.5 $4 / $20; Claude Sonnet 5.5 $2 / $10; Claude Haiku 4.5 $1 / $5.
Is my text sent anywhere?
No. The count runs in your browser and your text is never uploaded.
How Claude counts tokens
Claude uses its own tokenizer, which Anthropic does not publish as a library. In our own test (the same prompt in 41 languages, counted with each provider's token-counting API, October 2026), Claude Opus, Sonnet and Haiku 5.5 used 38% more tokens than GPT for the English version, and more for most other languages: 82% more in Korean, 73% in Spanish, 16% in Japanese. This counter applies the measured figure for the language you paste, so the Claude numbers above are an estimate, not an exact count.
Need the exact count for a request? Anthropic's API has a token counting endpoint (count_tokens) that returns the precise number before you send. Use this page for quick estimates and comparing models; use the API when you need the exact figure.
Claude API prices
| Model | Input / output per 1M | 1,000 chat requests |
|---|---|---|
| Claude Fable 5.2 | $8 / $40 | $38.64 |
| Claude Fable 5.1 (legacy) | $10 / $50 | $48.30 |
| Claude Opus 5.5 | $4 / $20 | $19.32 |
| Claude Sonnet 5.5 | $2 / $10 | $9.66 |
| Claude Haiku 4.5 | $1 / $5 | $3.60 |
Chat request = 1,500 input + 400 output tokens of English text, after Claude's measured tokenizer difference (+38% for English). Standard list prices, no caching or batch discounts. (2026-10-14)