Costs

Athena is paid for from your credit balance. You pay for the AI work each request does, and the machine and services Athena creates are billed like any others.

What a request costs

Athena's AI usage is charged to your balance as it works, like everything else on Appentic. Each finished request shows what it cost and how long it took under its last message, and your balance on the Billing page is always up to date. Troubleshooting requests are charged the same way.

What a request costs depends on how much Athena has to read and write. A first build costs more than a small change, and a big change costs more than a small one.

Type /cost in the box to see:

• what the project costs a month to run, part by part
• what Athena has cost on the project so far, and what the last request cost
• your balance, and the balance at which you're emailed a low-balance alert

Waiting for your answers costs nothing. Commands the page answers itself, such as /status, /cost and /undo, cost nothing either. Pictures add a little to the cost of the request they come with.

Machine and services

The machine and services Athena creates cost the same as ones you create yourself. You pay the machine's monthly price plus a monthly fee per service, charged by the second. See How billing works. Type /cost to see what your app's machine and services cost a month.

In Auto, the machine starts, and is billed, from your first request, so it's ready by the time the code is. Manual asks first, and Plan never starts one. The machine and services keep running after the request ends, because that's what keeps your app online. To stop paying for them, delete the project.

Athena adds things with a fee of their own, such as backups, a firewall or a volume, only when you ask. The commands for them say what they cost first, with a button to go ahead.

Spending limits

Athena never spends the credit kept back to keep your projects running (your suspension reserve). It can't start a request unless your balance is above it, and it stops a request that reaches it. Add credits and ask again to carry on.

A single request also has a spending limit set by Appentic, well above what a normal request costs. It's there to stop a request that has gone wrong from running up a bill.

Keeping costs down

• Be specific. "Add a notes field to each delivery" costs less than "improve deliveries", because Athena spends less working out what you mean.
• One feature per request. Smaller requests read less and finish sooner.
• Plan big changes first. In Plan mode, Athena looks into a change before building it, so you can correct the plan before paying for the build.
• Show, don't describe. A screenshot of the layout you want saves Athena guessing, and you correcting.
• Stop a request that's going the wrong way. Click Stop in the box, then ask again with more detail. You pay only for what it used.