Frequently asked questions
- How is reading time calculated?
- Reading time is based on an average reading speed of 275 words per minute — no bonus points for skimming.
- How is speaking time calculated?
- Speaking time is based on an average speaking speed of 180 words per minute. Auctioneers are not the average.
- What is the reading level score?
- Reading level is based on the Dale–Chall readability formula, which scores text using sentence length and the proportion of words outside a list of ~3,000 words familiar to most 4th-grade readers. It needs at least two sentences to produce a score — and no, Howy won't judge your vocabulary, he can barely spell "Dale–Chall" himself.
- Can I work on more than one piece of text at a time?
- Yes — use the tabs above the text box to open as many separate texts as you need, like browser tabs but for your unfinished drafts.
- Can I count just part of my text?
- Yes — select (highlight) any portion of the text and the "Selected" count updates to show the word count for just that suspiciously long paragraph.
- Does it work with languages other than English?
- Word and sentence counting uses your browser's built-in Unicode text segmentation, which handles a range of scripts reasonably well. Reading-level scoring (Dale–Chall) is English-only and shows "N/A" for other languages.
- Is my text stored or sent anywhere?
- No — all counting happens locally in your browser. Nothing you type is uploaded, saved, or read by us. Your diary entry is safe — even Howy isn't reading it, he's just here for the count (and maybe a nap).
- Who's Howy?
- Howy is our mascot — a cat who's green on the outside but 100% orange cat on the inside: chaotic, easily distracted, and weirdly good at this one specific thing.
How Many Words for Coders
- What models does the token counter support?
- Exact tokenization for GPT-4o / GPT-4o mini and GPT-3.5 Turbo, using the same tokenizers (
o200k_base and cl100k_base) those models use.
- Is the token count exact or estimated?
- Exact for the supported models — it runs the real tokenizer those models use in your browser, not an approximation based on word or character count.
- What's the estimated cost?
- An estimate of the input cost for your text, based on the selected model's published per-1M-token input price as of the date shown next to it — useful for ballparking, not a bill.
- What's the context window bar for?
- It shows how much of the selected model's context window your text uses — e.g. GPT-4o's 128,000-token limit — so you can see at a glance if you're getting close.
- Can I count just part of my code or prompt?
- Yes — select any portion of the text and the token count updates to show just that selection, same as the word counter's selection count.
- Is my code or prompt sent anywhere?
- No — tokenization runs entirely in your browser. Nothing you paste is uploaded, saved, or read by us.