| ๐ช Token estimate | Instant estimate plus characters, words and lines |
| ๐ธ Cost per tier | Input and output price per 1M tokens, times the number of calls |
| ๐ Context check | Tells you when prompt + expected output will not fit a model's window |
| โ๏ธ Editable prices | Rename tiers and set your own prices, saved in your browser |
| ๐ Private | Your text never leaves the page. No server, no tracking |
Open the live demo, or run it locally:
git clone https://github.com/CyberLoop-dev/tokenwise.git
cd tokenwise
open index.html # or just double-click itNo build step. No dependencies.
Real tokenizers differ from model to model, so tokenwise uses a transparent rule of thumb:
- about 4 characters per token for ASCII text
- about 1.5 characters per token for everything else (Persian, CJK, emoji, ...)
Treat the number as a good ballpark, not an invoice. For exact counts use your provider's own tokenizer or token-counting endpoint.
The three tiers that ship with the page are examples, not real price lists. Prices change often, so edit them to match your provider's pricing page. Your changes are stored in localStorage on your device.
MIT