Privacy Policy

• We do not use your prompts or model outputs to train any model, ours or anyone else's.
• We do not sell your data, and we do not share it with advertisers.
• We log request metadata — model, token counts, latency, status — because we cannot bill you or show you a usage history without it.
• We never see your password. Authentication is handled by Clerk.
• We never see your full card number. Payments are handled by our payment processor.

Account data. Email address, and the display name and avatar your identity provider gives us if you sign in with GitHub or Google. Held by Clerk, our authentication provider.

Billing data. Your credit balance, ledger entries and purchase history. Card details go directly to our payment processor and never touch our servers — we store only a processor reference, the last four digits and the card brand.

Request metadata. For every API call: timestamp, model, input and output token counts, latency, HTTP status, the API key used, and the computed cost. This is what your usage page is built from.

Technical data. IP address and user agent, used for rate limiting, abuse prevention and security investigation.

We do not use your prompts or the models’ outputs to train, fine-tune or evaluate any model. Not ours, and we do not pass them to anyone who does.

Prompt and output content is held only for as long as it takes to serve the request and return it to you. It is not written to our long-term stores and does not appear in your usage history — which shows what a request cost, not what it said.

Two honest exceptions. Content may persist transiently in operational logs during an active incident investigation, and asynchronous jobs such as video generation necessarily hold their inputs and results until you retrieve them, after which they are deleted on the schedule in §5.

Requests are served by upstream inference providers, who process content in order to generate a response. We contract for equivalent no-training treatment, but their handling is governed by their own terms.

Where the GDPR or a comparable law applies, we rely on: performance of our contract with you, for account and billing data; legitimate interests, for security, abuse prevention and service improvement; and legal obligation, for financial records we are required to retain.

We do not rely on consent for any of the above, which is why there is no consent banner demanding one. We set no advertising or cross-site tracking cookies — only the cookies required to keep you signed in.

• Request metadata: 24 months, then deleted.
• Asynchronous job inputs and results: 30 days from completion, or immediately on your request.
• Billing and ledger records: kept as long as tax and accounting law requires, typically seven years. These survive account deletion because we are not permitted to destroy them.
• Account data: until you delete your account, then within 30 days.
• Security logs: 12 months.

We use a small number of subprocessors, each for one job: Clerk (authentication), our payment processor (payments), Vercel (hosting), and the inference providers that serve model requests.

Every one is bound by a data processing agreement. None of them is permitted to use your data for their own purposes. We do not sell personal data, and we do not share it for cross-context behavioural advertising — under the CCPA’s definitions or any other.

Our infrastructure and our providers operate across multiple regions, so your data may be processed outside the country you are in. Where data leaves the UK or EEA we rely on the UK IDTA or the European Commission’s Standard Contractual Clauses.

Enterprise customers with regional processing requirements should contact affiliate@xark.io before onboarding rather than after.

Depending on where you live, you may have the right to access, correct, delete, port or restrict processing of your personal data, and to object to processing based on legitimate interests. Residents of California, Colorado, Connecticut, Virginia and comparable jurisdictions have equivalent rights, including the right not to be discriminated against for exercising them.

Most of this is self-service in the dashboard. For anything else, email affiliate@xark.io. We respond within 30 days and do not charge for it.

If you are in the UK or EEA you may also complain to your data protection authority. We would rather you came to us first, but that right does not depend on our preference.

Transport is TLS-encrypted end to end. API keys are stored hashed, not in plaintext, which is why we can only show you a key once at creation. Access to production systems is limited to staff who need it and is logged.

To report a vulnerability, email affiliate@xark.io. We will not pursue legal action against good-faith research that respects user privacy and does not degrade the service. See the compliance page for details.

The service is not directed at anyone under 18 and we do not knowingly collect their data. If you believe a child has given us personal data, email affiliate@xark.io and we will delete it.

If we change this policy materially — particularly anything in §3 — we will email your account address before it takes effect, not after.

Innoprise Incubator INC, operator of AI Token Router. affiliate@xark.io