Docs
← Back to TalkIDE
TalkIDE documentation · preview

Understanding AI credit and hosting billing

TalkIDE charges you in two separate ways: AI credit, which you pay for in advance, and hosting, which is billed after the fact based on what your published apps actually use. This page explains why it works that way, and what happens at the edges: running low on credit mid-task, or a hosting payment not going through.

Why there are two separate bills

AI credit and hosting are billed on opposite schedules on purpose. AI usage is something you actively spend as you work with the AI team, so a prepaid balance keeps you in control of it: you decide how much to add, and when. Hosting is more like a utility bill. It accrues quietly in the background as your published apps run, whether or not you’re actively working on them, so it’s billed the way utilities usually are: measured as it happens, invoiced after the month closes.

Every account starts with some AI credit

New accounts start with $10.00 in AI credit automatically. There’s nothing to set up and no card required to get going. That starting amount is what TalkIDE gives every new account today, though it’s a starting point rather than a fixed promise, so treat it as the current default rather than something guaranteed to stay exactly that forever.

Topping up when it runs low

When your balance gets low, you can add more AI credit yourself: minimum $5, maximum $500 per top-up, in whole dollars. It charges your saved card immediately through Stripe. Topping up is a one-time action, not a subscription: nothing auto-renews, and you decide each time whether to add more.

For the exact steps, saving a card for the first time, and where receipts show up, see Adding a payment method and topping up AI credit.

What “Usage limits” actually limits

It’s easy to assume AI credit has a dollar-amount spending limit you can configure, the way hosting does. It doesn’t. What AI credit does have is a Usage limits panel showing two rate windows, a 5-hour rolling window and a weekly rolling window, each shown as a percentage that resets on its own timer. Those are about how much you can use in a given stretch of time, not a money cap, and they’re not something you set yourself.

The only dollar-amount spending cap in TalkIDE today lives on the hosting side, covered below.

If your credit runs out mid-task

This is the part people worry about most, so it’s worth being precise: TalkIDE checks your AI credit balance before every individual step the AI takes, not just once at the start of a task. That means whatever step is already running finishes normally. The AI just can’t start the next step until you top up. Nothing gets cut off mid-response, and nothing is lost. If you run out partway through, the current step wraps up safely, and picking back up is as simple as adding more credit.

Hosting: pay for what your apps actually use

Hosting covers the compute, database, storage, and network your published apps use while they’re running. It’s metered hourly and invoiced once a month, on the 1st. There are no fixed tiers or quotas to pick between: you’re billed for what your apps actually consume.

An app with no active deployment, including an archived project, uses nothing and costs nothing. A preview deployment costs the same as any other published environment: both are just apps running, and both are metered the same way.

If a hosting payment doesn’t go through

If a hosting charge fails, TalkIDE doesn’t suspend anything on the spot. There’s a short automatic retry sequence over about a week first. Only if that doesn’t resolve things does your account enter a grace period, about a week long, before anything actually changes.

If the balance is still unresolved after that, your published apps are paused: scaled down, not deleted. Nothing about your project’s code or data is removed. Reactivating them is just a matter of clearing the outstanding balance.

Setting a monthly cap on hosting

If you’d rather not be surprised by a hosting invoice, you can set a monthly hosting spending cap in your billing settings: a simple dollar figure, with no limit set by default. It’s designed to stop new hosting usage from accruing past the amount you set.

Seeing what each environment costs

Once you’re signed in, your billing page shows hosting cost broken down by environment, not just one lump total. That’s useful if a project has more than one published environment and you want to see which one is actually costing money.

Was this page helpful?

Thanks for the feedback.