How Many Tokens Is My Text? (And How to Estimate API Cost)

If you're building with the OpenAI, Anthropic or Google APIs, your bill is measured in tokens — not words or characters. Knowing roughly how many tokens your prompt uses helps you predict cost and stay within context limits. Here's the quick version, plus a free token & cost calculator.

What is a token?

A token is a chunk of text — often a word or part of a word. Common rules of thumb for English:

• ~1 token ≈ 4 characters
• ~1 token ≈ 0.75 words (so 100 tokens ≈ 75 words)
• 1,000 tokens ≈ ~750 words ≈ a page and a half

Code, punctuation and non-English text tokenize differently, so these are estimates — use a real tokenizer for exact counts.

How to estimate API cost

Cost = (input tokens × input price) + (output tokens × output price). Prices are quoted per 1 million tokens, and output usually costs more than input. So:

• Estimate your input tokens (prompt + context)
• Estimate output tokens (the response length you expect)
• Multiply each by the model's price and add them up

🧮 Count tokens & compare prices →
Free Token Cost Calculator — paste text, see the token count and compare GPT / Claude / Gemini costs side by side.

Ways to cut token cost

• Trim system prompts and redundant context
• Cap output length with a max-tokens setting
• Use a smaller/cheaper model where quality allows
• Cache or reuse static context instead of re-sending it

FAQ

Is token count the same across models? No — different models use different tokenizers, so the same text can be a slightly different number of tokens.

Does output cost more than input? Usually yes — check each provider's per-million-token pricing.