> ## Knowledge Base Index
> Fetch the complete knowledge base index at: https://help.booksgpt.ai/sitemap.xml
> Use this file to discover available pages before exploring further.
> Pure-Markdown content can be obtained by appending a '.md' suffix to the content URLs listed in the sitemap (without the trailing slash).

# AI credits: what they are and what spends them

Credits are what the AI assistants spend. They apply to the Free and BooksGPT+ plans.

|| If you are on a lifetime plan or a trial, credits do not apply to you. Your AI use is not metered, and the credits page says so when you open it.

Open **All pages** in the left sidebar, then under **Plan** click **AI credits**.

## What spends a credit

* Asking **Copilot** a question about your books.
* Asking **Tax-GPT**.

## What never spends a credit

* Categorizing transactions. That always runs free.
* Bank feeds, imports, invoices, bills, reports and reconciliation. None of them are AI-metered.

| So the day-to-day bookkeeping that keeps your books current does not consume credits at all.

## Your balance

The big number is what you have left. Below it, the split:

* Credits from this month's plan, which reset at the start of each month. They do not roll over.
* Purchased credits, which do not expire.

![The AI credits balance showing the total, with this month's plan credits and purchased credits below](https://booksgpt.ai/help/credits/balance.png)

## Top up

If you run out before the month ends, buy a pack under **Top up**.

* Three sizes are offered. The larger packs give you more credits per dollar.
* Checkout is handled by Stripe, and the credits land as soon as it completes.
* Your monthly allowance is always spent first, so a top-up is only touched once this month's credits are gone.

## Where they went

**Recent activity** lists every change to your balance, with the date and whether it was Copilot or Tax-GPT.

![Recent activity listing monthly credits, usage and top-ups](https://booksgpt.ai/help/credits/activity.png)

## Good to know

* A longer question about more data costs more than a short one.
* Running low? A shorter, more specific question is usually cheaper and gives a better answer.