Skip to main content

How AI usage is billed

How you pay for AI in Studio depends on which plan you're on. This page covers both AI billing models.

Two AI billing models​

  • Managed plan ($0/mo) — pay-as-you-go. Studio uses platform API keys, and you're billed per token used. Top up your balance whenever you need to.
  • BYOK plan ($20/mo) — bring your own keys. You supply your own Anthropic and OpenAI API keys, pay those providers directly at their rates, and Studio charges a flat monthly subscription on top.

(The third plan, Dedicated, is about where your projects run, not how AI is billed — Dedicated users are still on either Managed or BYOK for AI. See Plans for the full picture.)

Most of this page is about the Managed plan. BYOK users — skip to BYOK at the bottom.

What tokens are​

AI models charge by token, a chunk of text roughly equal to ¾ of a word. Both the message you send (input) and the assistant's reply (output) count. Bigger requests, longer replies, and longer projects cost more.

Different models have different per-token rates — see Choosing an AI model for the trade-offs.

Your balance​

Your balance shows in the top bar of Studio as a small badge with a lightning-bolt icon: $X.XX USD. It updates after each AI turn, so you see the cost of working in close to real time.

The badge changes appearance as your balance drops:

  • Normal — comfortable balance.
  • Low balance (gold, pulsing) — between $0.01 and $2.00 remaining. A warning banner also appears in the chat area suggesting you top up.
  • Depleted (red) — your balance has run out. The AI stops mid-task and a Balance Depleted overlay appears with a TOP UP NOW button. Topping up resumes the conversation where you left off.

Topping up​

Select the balance badge (or the TOP UP button on a low-balance warning) to open the top-up dialog. Pick a preset — $10, $25, $50, or $100 — or enter a custom amount (minimum $10). Enter your card details and confirm; payment is processed through Stripe.

Your balance updates as soon as the payment succeeds.

Per-conversation usage​

Near the balance badge there's a small running readout for the current chat, in the form:

$0.34 · 12.4k in / 3.1k out · 5 turns

It shows what this conversation has cost you so far, how many input and output tokens you've used, and how many AI turns have happened. It's there so you can see what's expensive and what isn't.

Billing history​

Every charge, top-up, and refund is recorded. Open Settings → Billing & Usage to see the full history with timestamps and amounts.

Tips for keeping costs down​

  • Pick the right model for the job. A small copy tweak doesn't need Opus. See Choosing an AI model.
  • Be specific. Clear, focused requests run shorter — and produce better results — than vague ones.
  • Break large requests up. One enormous "rebuild the whole site" prompt costs more than a series of targeted changes, and you get to review along the way.

BYOK plan​

On BYOK, the picture is simpler:

  • You pay a flat $20/mo to Studio (billed by Stripe).
  • The AI runs on your API keys (one for Anthropic, one for OpenAI), which you set up in Settings → API Keys. Token costs go directly to those providers at their rates, on their bills — Studio doesn't charge for usage.
  • The top bar shows a BYOK badge instead of a dollar balance. Per-conversation readouts show token counts only, no dollar amount.
  • DeepSeek isn't available on BYOK.

For instructions on setting up your keys, see Using your own API keys (BYOK).

Next​