Pricing & Billing

Simple usage-based pricing for Aivana MMI, paid from prepaid credit.

Aivana MMI uses simple usage-based pricing. You pay for the tokens processed by your requests and generated in your responses.

A request with 5,000 input tokens and 2,000 output tokens, with no cached input, is charged:

• Cached input: the part of your input a run verifiably reuses from cache, such as earlier turns and files a follow-up sends again, is charged at the cached-input rate. Everything else is standard input.
• Files: attached files are billed as input tokens. See Files & Documents for how each kind is counted.
• Web search: a request that searches the web is charged every token used to answer it. See Web Search.
• Stopped requests: if you stop an answer partway, you're charged for your prompt and the part you received.
• Failures: a request that fails, or is cut short, on Aivana's side isn't charged.

• Aivana is prepaid. You buy usage credit in AI Studio and each request draws down your balance.
• Promotional tokens, when you have them, are spent first. Purchased credit is spent after.
• Promotional tokens and purchased credit expire. Promotional tokens expire 60 days after you receive them, and purchased credit one year after purchase; the expiry dates of your promotional tokens are shown on your Billing page.
• Your remaining balance is shown on the Billing page in AI Studio.
• When your balance runs out, requests are refused with a 402 (insufficient_credit) until you add credit.

You can save a card on the Billing page, or tick "save this card" when you add credit, so the next purchase is one click. Card details are entered in a secure form provided by Stripe, our payment processor, and never reach Aivana's servers; Billing shows only the card's brand, last four digits and expiry. A saved card is charged only when you add credit or approve auto-reload. You can make a card the default or remove it at any time.

A bug in your code can send requests in a loop and spend a whole balance in an afternoon. You can cap how much purchased credit your account, or a single API key, may spend in a day, a week or a month (UTC; Monthly is the first choice offered) on the Billing and API Keys pages. When the limit is reached, requests are paused until the period starts again, or until you raise or remove the limit. You can also see how much of the limit you have used on the Billing page, and the sidebar warns you as you approach it. Usage paid for with promotional tokens is never limited and does not count. An account limit counts what you have already spent from purchased credit in the current period, including before you set it, so a limit set below what you have spent this month pauses requests straight away. A key's limit counts from the moment you set it. The limit is yours alone: Aivana never sets or changes it.

Use Usage & Insights to see token usage, billed usage, run history and balances.

Usage & Insights can compare your actual Aivana cost with an estimate of what the same workload would cost at an AI benchmark's published pricing.