If you've run into a "message too long" error in ChatGPT or Claude, or you're trying to figure out whether your document will fit inside a context window, it helps to actually understand what a token is instead of just guessing.

What is a token, really?

A token is a chunk of text the model processes - sometimes a whole word, sometimes part of a word, sometimes just punctuation. It's not the same as a character or a word, which is exactly why "how many tokens is 1000 words" doesn't have one fixed answer - it varies by language and by how the specific model splits text, but a commonly used rough estimate for English is about 750 words per 1,000 tokens, or roughly 1.3 tokens per word.

Context window, in plain terms

The context window is the total amount of text (counted in tokens) a model can "see" at once - your prompt, any attached files, and its own response all count against that same limit. Different models and plans have different context window sizes, and that number has been growing steadily across new model releases, so a conversation or document that wouldn't fit a year or two ago often fits comfortably now.

Why ChatGPT or Claude sometimes cuts off mid-response

If a conversation gets long enough to approach the context limit, older messages may get dropped from what the model can "remember," or a single very long response can get cut off mid-sentence. Starting a fresh conversation, or asking the model to continue from where it stopped, is usually the simplest fix.

A practical substitute for exact token counting

Dedicated token counters for each specific model exist, but for a quick everyday estimate, a word and character count gets you most of the way there - roughly divide your word count by 0.75 to estimate tokens. Our Word Counter gives you that word and character count instantly as you paste or type, which is a close enough proxy for most everyday planning.

If you're pasting AI output somewhere and the formatting comes along messily (stray asterisks, broken markdown), our guide to cleaning up ChatGPT and Claude formatting covers that separately.