VAKS · Effective September 14, 2026

Privacy policy

This policy explains how VAKS processes inference content when providing model inference services.

Operator

The service is operated by Dharma LLC. Privacy questions may be sent to support@vaks.ai.

Inference content

Vaks processes prompts, message content, tool definitions, and other request content submitted to the inference API, together with model-generated responses, solely to return the requested inference result.

Prompts are not logged. Responses are not logged. Inference content is not used to train models.

Transient caching

The inference runtime may maintain transient in-memory prefix or key-value caches to accelerate active and repeated computation. These caches are runtime memory, are not a persistent prompt or response log, and are cleared through normal workload and infrastructure lifecycle events.

Processing location

Service processing occurs in Belgium and the Netherlands. Vaks is designed for EU-resident routing.

Infrastructure providers may process data on our behalf under contractual data-protection and security obligations.

Legal disclosure

Vaks may preserve or disclose information when required by applicable law, valid legal process, or a binding request from a competent authority. Where legally permitted, requests are evaluated for scope and validity.

Your rights

Depending on applicable law, individuals may request access, correction, deletion, restriction, or portability of personal data, or object to processing. Send requests to support@vaks.ai. Because prompts and responses are not logged, Vaks may not hold inference content after a request completes.

Data policy

A concise provider disclosure is available on the Vaks data policy page.

We may update this policy by posting a revised version on this page and changing the effective date. Material changes will be identified in the revised notice.