Seeing Where Your Credits Went

Playcode keeps a receipt for every AI request you paid for. This article shows you where that record lives, how to read a single expensive message, and which spending controls actually stop something.

Open your usage history

Usage is a page inside Settings.

1. Open the workspace menu in the left sidebar and choose Settings…
2. In the settings sidebar, under Workspace, click Usage

The page opens on Today. The other periods are Yesterday, This Week, This Month, Last Month, This Year and All Time. Picking one reloads the whole page.

The numbers fill in for the workspace owner and its admins. If you are a plain member of someone else's workspace, the page tells you who the AI usage is billed to instead of showing their spend.

What the Usage page shows

The page has four parts, top to bottom.

• The summary band. Credits used in the selected period, the number of AI Messages, and Active members. The members tile appears only when the workspace has more than one person in it.
• Spending by service. A grouped table: AI (chat and agents) with its message count, Email when this billing cycle metered any sends, and one row per cloud app with Compute, Disk and Traffic broken out under it. The AI group follows the period you picked. The cloud rows always cover the current month.
• By member. Credits and request count per person, with the date they last ran something.
• Recent Activity. The request-by-request list, newest first, with a Load more button and a "Showing X of Y" count under it.

Read one expensive message

Each row in Recent Activity is one AI request: the model that ran, who ran it, what kind of work it was, the time, then the credits and the token count. Click a row to open its breakdown.

A bar shows the proportions, then one line per part with its tokens and its credits:

• Input - what the AI read this time.
• Output - what the AI wrote back.
• Cache write - the one-time cost of storing your prompt and project context so later messages can reuse it.
• Cache read - reused context, charged at a fraction of a fresh read.

Under the lines, Playcode prints a plain note saying which part dominated, plus how long the request took. Records written before per-token cache detail existed say so instead of guessing.

One message you send can produce several rows. The agent works in steps, and every step is its own request with its own row. That is why the Messages count is normally higher than the number of things you typed.

Find the chat that spent the credits

The Usage page does not group by chat. For one chat, open the project menu at the top of the editor, the one with the project name on it. The foot of that menu shows two lines for the chat you have open:

• Spent in this chat - credits this chat has used so far.
• Context - how full the conversation is, in tokens and as a percentage.

Past about 80% full, the context reading turns amber and the menu tells you to start a new chat for the next task. That is the moment a chat starts getting expensive, and it is the cheapest warning you will get.

What makes one message cost more than another

• The build level. Economy is the low-cost level and handles simple edits, fixes and basic pages. Standard is the premium level. Ultra is the most capable and the highest cost. Free accounts run on Economy only.
• Project size. The bigger the project, the more files the AI reads before it can change anything, on every single message.
• Conversation length. Every message carries the conversation so far along with it, so late messages in a long chat cost more than the first one did.
• Your AI defaults. Settings holds a default build quality, an effort setting for Standard, and a context size. Playcode's own hint on the context setting says it plainly: a larger context can cost more.

For what to do about all this, read How to Save Credits.

Credits are not only for chat

Cloud apps and Playcode Email draw on the same credits as the AI. A sleeping cloud app still bills for its disk, while the hours meter runs only while it is awake. Email sends cost credits only past the allowance included in your plan. That is why the Spending by service table exists at all: a month where you barely chatted can still show a bill.

Playcode also protects part of your balance for your computers. AI work cannot spend that reserve, so a heavy build session cannot quietly take your live site offline.

What spending controls actually do today

The monthly notice - it does not stop anything

On Pro or Max, Settings → Billing & plans has a Spending controls section with one field: credits in a month. When the workspace passes that number, Playcode emails you once for that month. The dialog states it exactly: a heads-up only, nothing stops.

So treat it as a smoke alarm, not a circuit breaker. Your chats keep running, your apps stay online, and the credits keep being spent after the email lands. Leave the field empty to turn the email off. Change the number and the notice re-arms for the current month.

The auto-recharge monthly maximum - this one is a real cap

The one hard money number you hold is in auto-recharge. Its optional monthly maximum caps what Playcode may charge your card in automatic top-ups per month. Once the month's automatic top-ups reach it, no more are taken. It caps charges, not usage. See How Auto-Recharge Works.

Running out of credits

There is no switch that says "hard cap my AI at N credits this month". The real floor is your balance: when a workspace has no credits left, AI requests are refused rather than run on credit. If you want that floor to hold, keep auto-recharge off and top up by hand.

Where to go next

• How Credits Work - what a credit buys and why no honest plan is unlimited.
• How to Save Credits - the habits that cut a bill in half.
• AI Credits Balance - where credits come from, the order they are spent in, and when they expire.
• How Auto-Recharge Works - thresholds, top-up amounts, and the monthly maximum.

Need help?

If a row in your usage history does not make sense, reach out through the chat button in the bottom right corner of any Playcode page. A human will read it with you.