LLMCost
Free developer tool

Token Counter & Cost Estimator

Paste text, see the token count instantly — then see what that request costs on every model, ranked cheapest first. 100% client-side: your text never leaves the browser.

Tokenizer:
0
tokens
0
words
0
characters

Exact count via OpenAI's o200k_base tokenizer — the same encoding GPT-5.x/4o models use. Claude and Gemini use private tokenizers; their counts typically land within ±10% of this estimate.

Token rules of thumb

  • 1 token ≈ 4 characters ≈ 0.75 words of English
  • 1,000 tokens ≈ 750 words ≈ 3 pages double-spaced
  • Code tokenizes less efficiently: expect ~3 chars/token
  • Non-English text (CJK especially) can use 2-3× more tokens per word
  • A 200K context window ≈ a 500-page book in a single request