Every AI model has a limit on how much text it can handle at once. The Token Counter shows you how close you are to that limit — for each model — so you can adjust before you send your prompt. No more "prompt too long" surprises.
The token counter appears in the Prompt Composer toolbar and in the Document Editor status bar. It's always there when you're writing.
What you'll see
The counter
A count for each AI model you use. Color indicators tell you where you stand: green (good to go), yellow (getting close), red (over the limit).
Model comparison
See how your prompt fits across all your models at a glance. Helps you decide which model to use, or what to cut for a smaller one.
How to use it
Check while you write
Look at the status bar at the bottom of the editor — it updates as you type.
Compare across models
In the Composer, the token bar shows counts for all your models side by side.
Trim if needed
If you're over the limit, try shortening a context card, removing one you don't need this time, or simplifying your prompt.
Good to know
A token is roughly ¾ of a word in English. Code and structured text tend to use more tokens per word than regular writing. Don't worry about memorizing this — the counter does the math for you.