Terms of service.

These terms are an agreement between you and Kottos AI, Inc. ("Kottos AI", "we") about the self-serve Kottos AI service: the gateway at api.kottos.ai and your account at data.kottos.ai. By creating a token or sending requests through the gateway, you agree to these terms, to the privacy policy and to the disclaimer. If you accept them for a company, you confirm that you can bind it, and "you" means the company. If your company has signed a separate agreement with us, that agreement takes precedence where the two differ.

The service

The gateway forwards your requests to AI providers with your own provider keys, routes each request by the route you choose, and records metadata about it. Your account shows your balance, usage, cost comparisons and daily request files. Kottos AI does not resell model access: each provider bills you directly for its tokens, and we charge a fee for the gateway.

Your account and token

• You must be 18 or older and sign in with an email address you control.
• Your Kottos AI token is shown once. Anyone who has it can send requests on your account and spend your credit, so keep it secret. If it leaks, replace it on your account page; the old token stops working within about a minute.
• You are responsible for everything sent with your token.

Your provider accounts

• You need your own account and key with each provider you use, and you must follow that provider's terms and usage policies. We are not a party to your agreement with a provider.
• Each provider charges you for what it processes, including requests that fail, requests retried on another provider, and requests a hedging route races across two providers. We do not pay provider charges and cannot refund them.
• We are not responsible for a provider's outages, prices or changes, its decisions about your account, or what its models produce.

Routes

Each request is served by the route its x-kottos-step header names, or by your default route. A route decides which providers and models can serve a request, and it can set the service tier, speed or region. Some routes do more than forward: they race two providers and keep the first answer, retry on another provider after a failure, or move a conversation to a different model under a rule. The account page describes what each route does. We may add, change or retire routes, providers and models, and a provider can retire a model at any time.

Fees and credit

• The fee. $0.05 per million input, cache-write and output tokens, and $0.005 per million cache reads, each capped at 4% of the provider's list price for that token. A refused or failed request carries no fee, and on a hedged request only the leg that served you does.
• Prepaid credit. Fees are paid from your Kottos AI credit and taken about once an hour from our request records. Because of that delay your balance can briefly go below zero, and a negative balance comes off your next top-up.
• At zero. When your balance reaches zero, your token pauses: the gateway refuses its requests until you add credit, and it works again within about a minute of the payment.
• Starting credit. A new account gets $10.00 of starting credit, once per email address. It pays fees only, has no cash value, and cannot be withdrawn or transferred. We may change or end the starting credit for new accounts.
• Top-ups. You add credit by card through Stripe, from $5.00 to $2,000.00 per payment, in US dollars. If you pay in another currency, you are credited the US dollar amount Stripe reports. Any tax shown at checkout is not added to your credit.
• What credit pays for. Kottos AI fees only, never provider charges. Credit does not expire while your account is open.
• Refunds. Unused credits are refunded on request within 30 days of purchase: email [email protected]. Credits already spent on fees are not refunded. The starting credit has no cash value.
• Fee changes. We will email you at least 14 days before a fee increase takes effect, and a change applies only to requests sent after it takes effect.

Acceptable use

You may not:

• break the law, or a provider's terms or usage policies;
• send data you have no right to send, including other people's personal information;
• send protected health information covered by HIPAA through a self-serve account, since we sign no business associate agreement for one;
• get around rate limits (currently 60 requests a minute per account), credit checks or authentication, or create accounts to collect starting credit;
• probe, attack or overload the gateway or our pages, or use them to attack anyone else;
• resell access to the gateway, or share your token outside your company;
• use Kottos AI if you are on a US sanctions list or in a country under a US embargo.

We may suspend or close an account that breaks these rules, and we will tell you why unless the law or the security of the service prevents it. Security research on your own account is welcome: report what you find to [email protected] and leave other people's data alone.

Your content and records

• As between you and us, you own your prompts and the replies you receive. We process them only to forward them, and we do not store their text.
• You allow us to use the records of your requests to run, secure, bill and improve the service, and, combined with other accounts' records, to measure providers, improve routing and publish statistics that do not identify you, as the privacy policy describes.
• If you send us suggestions, we may use them without owing you anything.

Estimates on your account

Cost comparisons, savings, simulations and timing estimates on your account page come from list prices and our own measurements. They are estimates, not invoices or promises, as the disclaimer explains. Your provider's invoice decides what you owe the provider, and our request records decide our fee.

Availability and changes

We work to keep the gateway available, but the self-serve service has no uptime commitment or service level. We may change or stop features. If we shut the self-serve service down, we will give at least 30 days' notice by email and refund unused purchased credit.

Open source

The gateway is built on llmbridge, which is open source under the Apache License 2.0. These terms cover the hosted service only and do not change that license.

Ending

You can stop using Kottos AI at any time and ask us to close your account. We may close an account that breaks these terms. If we close your account for any other reason, we will refund unused purchased credit.

No warranty

Kottos AI is provided "as is" and "as available". To the extent the law allows, we disclaim all warranties, including merchantability, fitness for a particular purpose and non-infringement. We do not promise that the gateway will be uninterrupted or error-free, that any provider or model will stay available, or that a route will lower your cost or latency.

Limit of liability

To the extent the law allows, we are not liable for indirect, incidental, special or consequential losses, for lost profits, revenue or data, or for charges from your providers. Our total liability for any claim is limited to the amount you paid us in the 12 months before the claim.

Indemnity

You will defend and indemnify Kottos AI against third-party claims arising from what you send through the gateway, from your use of a provider, or from your breach of these terms.

Changes to these terms

We may change these terms. We will update the date above and, for a material change, email account holders at least 14 days before it takes effect. If you keep using Kottos AI after that, the new terms apply.

Governing law

These terms are governed by the laws of the State of Texas, without regard to its conflict-of-law rules. Disputes go to the state or federal courts located in Travis County, Texas.

Contact

Kottos AI, Inc. · [email protected] · security issues to [email protected]