What a Message Costs

The three places you meet the price in the Code Studio: the cost under each turn's answer, the token meter under the box, and the popup when you hit zero.

Where the price shows up

Nothing in the Code Studio asks you to confirm a chat message — you type, you send, and the credits go. So the price is shown after the fact, in three places. They are worth knowing before you spend a long afternoon in one project.

Under the answer — the cost label

Under the AI's answer sits a small footer: the time it was sent, a badge with the model that answered, and the cost. The cost is one figure for the whole turn — everything your message set off: the plan, the steps that wrote the files, any repairs and continuations, and whatever was still settling after you pressed Stop. It is read from the billing ledger and sits under the last answer of the turn. It reads in credits — 11,6 credits, say. When the turn carries a step-by-step breakdown, the label adds a count such as · 3 phases, and hovering it opens a tooltip headed Where this turn's tokens went, step by step, with Charged at the bottom. Without that breakdown you see the plain figure, and its tooltip just reads Context this turn (input + output).

The label is drawn while Show Token Usage is on in Profile → Preferences → AI Assistant — it is on by default. A dash (—) under an earlier answer of the same turn does not mean that answer was free: its cost is in the figure under the turn's last answer, so nothing is counted twice. A dash can also mean nothing was recorded for that message. The project icon and store-description generations are not included in these figures.

The ⋯ menu — Message details

The ⋯ button at the end of the same footer opens Message details: the Model, what was Charged, and when it was Sent — the same facts, one tap away even with the label switched off. Under them is Copy message. The 👍 / 👎 buttons and the read-aloud speaker sit in the same row.

Under the box — the token meter

The AI is charged by the token, and the biggest part of a message is not what you type but what goes with it: your files and the recent conversation. So the strip under the message box carries an estimate of how much the next message will send. It is silent in a small conversation and starts reading ~12K tokens once the estimate passes ten thousand.

As the context fills up, the label changes:

at 70% it adds the percentage — ~90K tokens (70%) — and changes colour
at 90% it turns red and reads near limit!

A bar above the chat says the same thing in words: Context is 72% full, then Context 91% — older messages auto-compact, you can keep going. That last sentence is the point: you are not cut off at the limit. The older part of the conversation is folded into a summary so the newest messages still fit.

It is an estimate, not a bill

The meter is worked out in your browser from the newest messages, a capped share of your files and the smallest context window among the models you could switch to. It is there to tell you when a message is getting heavy, not what it will cost to the credit. The number you actually paid is the label under the turn's last answer once it arrives.

"You're out of credits" — the popup

The moment your balance reaches 0, a popup opens over whatever you were doing — though never while a brand-new account's welcome credits are still on their way:

> You're out of credits > Your balance hit 0 credits, so generating is paused. Your project is saved — top up and pick it straight back up.

The first row is Buy … credits — $…, with a real pack on it, usually the Standard pack. Its label is the same as that pack's Buy button on the Pricing page, price included, so the two always agree — and that price includes 21% VAT; the line under it says One payment, including 21% VAT. That is exactly what you pay, wherever you are: the VAT is already inside a pack's price (see Payment Methods & VAT). Under that sits the refund promise: if what VyroAI built does not run, the pack is refunded in full within 14 days — ask through the contact form on /support. The row takes you to the Pricing page. If the prices could not be loaded just then, the row simply reads Buy credits. If you still have welcome credits to claim, a second row offers those. The last row is Leave feedback, with a tag saying what a report can earn once someone here has read it. Maybe later closes it.

If a discount is running, the buy row carries an extra line under the price: the code in an orange chip you tap to copy, the deal in words, and the date it ends. The discounted price is not printed there, because the code is applied at the checkout.

It shows once each time you run dry — not again on every click while you stay at zero — and it comes back the next time your balance drops to zero after having been above it. If credits arrive while it is open — welcome credits landing late, a purchase in another tab — it closes by itself.

Nothing that was already running is thrown away at zero: the job finishes on what you have and you land on 0, never below. Running Out of Credits has the detail, and The "Large request" Warning is the one place the Code Studio does stop you before an expensive message.

Where to next

Next in Account & Billing: Why Are My Credits on Hold?
New to credits: How Credits Work — Why a message costs anything at all.
Change what you see: AI Assistant Preferences — The switch that shows or hides the label under each answer.
Before a big ask: The "Large request" Warning — The warning that stops an expensive message before it goes.
Down to your last few: Running Out of Credits — What zero actually means for a job already running.

Also linked from this page: Usage & History — the full ledger, once you want more than one message at a time

13 of 31 in Account & Billing · Last checked 23 September 2026 · Previous: Running Out of Credits