Docs menu

Docs / More

FAQ

Short answers about Cent AI, payment, budget packs, balance, privacy, limits and refunds.

Last updated


What is Cent AI?

Cent AI is a prepaid AI API. One key, cai_<secret>, reaches many models through three API formats on one host: OpenAI Chat Completions and OpenAI Responses at https://api.cent-ai.dev/v1, and Anthropic Messages at https://api.cent-ai.dev. You pay per token from a USD balance, and every response tells you what it cost. Start with the Quickstart.

Does it work with the OpenAI and Anthropic SDKs?

Yes. Change the base URL and the API key; the rest of your code stays the same. See the guides for the OpenAI Python, OpenAI JavaScript, Anthropic Python and Anthropic TypeScript SDKs. The key works in Authorization: Bearer and in x-api-key.

Which models can I use?

Chat models, and image models on the image endpoints. The current list, with the four rates of each model, is on the models page and the pricing page, and through GET /v1/models. The image endpoints are described on Images. Video, speech, transcription and embeddings endpoints are not open yet; they return 501 not_available_yet and are not charged.

How am I charged?

Per token, from your USD balance. Each model has four rates: input, cache read, cache write and output. Cache writes are billed at the input rate. The cost of a request is rounded up once, to $0.000000001. Cache rates apply only when the model reports cache tokens. See Pricing.

Is there a minimum charge per request?

Yes. Each successful request costs at least $0.00002, even when its tokens cost less. A model can have its own minimum, shown on its page. A request rejected before it reaches a model, for example with 401, 402 or 429, costs nothing.

Where do I see what a request cost?

In the usage.cost field of the response, in USD with 9 decimal places. When streaming, it is in the last usage chunk on Chat Completions (with stream_options.include_usage: true), in the last message_delta on Messages and in response.completed on Responses. Every request is also on the Requests page of the dashboard. See Request cost.

How do I pay?

With QRIS in rupiah, or with USDT or USDC on Solana. The minimum top-up is $1. Cent AI pays the payment processing fees for QRIS; for crypto, network fees charged by your own wallet are yours. Every payment is made by hand: there is no card on file. See Top-up.

Budget pack or pay-as-you-go?

Both use the same four per-token rates and the same rate limit.

Budget packPay-as-you-go
Price$1.99, $9.99 or $99, one-timeAny amount from $1
BudgetEqual to the priceEqual to the top-up
ModelsThe models included in that pack size; the $99 pack covers every chat modelAll models
RenewalPaid by hand, no automatic renewalPaid by hand
ExpiryNone while your account is activeNone while your account exists

Token figures shown for a pack, such as "up to N million tokens", are estimates. Charges are taken from promotional credit first, then pack budget, then pay-as-you-go. See Balance.

What happens when my balance runs out?

New requests return HTTP 402 insufficient_balance and are not sent to a model, and the dashboard shows a banner. Cent AI does not send low-balance emails. Requests already running finish and are charged in full, so the pay-as-you-go balance can go below zero; the next top-up offsets it.

Does my balance expire?

Pay-as-you-go balance does not expire while your account exists, and pack budget does not expire while your account is active. Promotional credit, such as the signup credit and referral rewards, expires 30 days after it is granted.

Is there credit to try it?

New accounts get a one-time promotional credit after signing in with a verified email. It works on every model for 30 days. The amount is on the Balance page.

Are my prompts stored?

No, unless you turn on 24-hour content storage for a key. Stored content is encrypted and deleted after 24 hours, and you can see it on the Requests page. Request metadata, such as time, model, tokens, cost and status, is always recorded. Your content is not used for training and is not sold. To produce a response, your request is sent to the third-party model provider that serves the model you chose. See the Privacy Policy.

What are the rate limits?

600 requests per minute per API key. The Playground has its own 600 RPM per account, separate from key limits, and draws on the same balance. Budget packs have the same limit. You can ask support to raise a key's limit. See Rate limits.

How do I cap my spending?

Each key can have a daily and a monthly limit, an IP allowlist and a model allowlist, and each project can have a daily or monthly budget. All are optional, and new keys have no limits. See Authentication.

Can I get a refund?

Unused balance is not refunded to cash. The full rules are in the Refund & Credit Policy.

Is there an SLA?

There is no contractual SLA. If a model fails before it sends the first byte, Cent AI retries the request on another route for the same model. See 5xx errors.

How do I contact support?

Email support@cent-ai.dev. Include the x-request-id of the response when you ask about a request, and never send your API key.