Grape AI comes with an allowance rather than a meter you have to watch, but when a reply is refused it helps to see where you stand.
Opening it
Open Settings, go to the Grape Pro tab and open the usage dialog from there.
The four bars
- Chat covers the assistant, including the work it does editing your notes
- Transcription covers turning recordings into text, with or without speaker identification
- Brainrots counts the videos you generated
- Infographics counts the posters you generated
- Slide deck images counts the pictures drawn for your slide decks. The slides themselves are written by the chat model and count under AI chat
Each bar is a percentage rather than a count, because six percent says what you want to know and 124 of 1500 makes you do the division. A few smaller caps are metered on the server and deliberately left out of the dialog: they exist to catch a runaway loop rather than to be budgeted against.
The two periods
The dialog opens on a five hour window, the limit people actually run into and the only one whose reset is close enough to wait for. Switch to the month for the longer view. A month here is only three to five windows wide, so it can genuinely run out first.
When you run out
Wait for the window to reset, which the dialog counts down to, or use your own API key or a local model instead. Keys you bring yourself are billed by that provider and are not metered here.
The colours
A bar turns amber past seventy percent and red past ninety, so a glance is enough.