How the token counter works
Language models don't read characters or words; they read tokens, chunks of text that are often a word, part of a word or a punctuation mark. API pricing and context limits are both measured in tokens, so the token count of your text drives both cost and fit.
This page counts tokens with o200k_base, the open tokenizer OpenAI publishes for its recent models, running locally in a Web Worker so typing stays fast even on long documents. Until the tokenizer has loaded, you see a quick estimate based on about four characters per token.
Prompt cost
Cost per request = input tokens × input price + output tokens × output price, with prices quoted per 1 million tokens. Where a vendor charges more for long prompts (for example above 272K or 200K tokens), the tier that matches your prompt length is applied automatically.
Prices come from our verified dataset (see the full LLM API pricing table). Introductory prices are excluded unless you switch them on, and if we could not confirm a price on an official page the table says "Price not verified" instead of showing a number. To estimate a month of traffic rather than one prompt, use the LLM API cost calculator.
Context window check
The context window is the maximum number of tokens a model can handle in one request, counting your input and its output. The table shows input + expected output as a share of each model's published window. Some vendors don't publish a window for every model; those show "Not published".
Tokens to words
For typical English prose, 1 token is roughly ¾ of a word, so 1,000 tokens ≈ 750 words. Code, numbers, non-English text and unusual formatting usually need more tokens per word; Japanese and other CJK text often needs about one token per character or more. Paste a sample of your real text above for an accurate ratio.