Data Policy
This policy describes how MADD handles the content and metadata of requests sent to the MADD LLM API at llm.madd.im, including requests routed to us by OpenRouter.
1. Zero retention of prompts and completions
We do not store prompts, input files, or model outputs. Request and response content exists only in memory for the time needed to generate the response and is discarded once the response is delivered. We do not write content to logs, databases, or backups.
2. No training
We never use prompts, completions, or any other customer content to train, fine-tune, or evaluate models, and we do not share content with third parties for such purposes.
3. Metadata we keep
For billing, capacity planning, and abuse prevention we record only the following per request:
| Field | Purpose |
|---|---|
| Timestamp | Billing period, uptime statistics |
| Hashed API key identifier | Assigning usage to a customer account |
| Model identifier | Billing per model |
| Prompt, cached, and completion token counts | Billing |
| HTTP status and latency | Reliability monitoring |
Metadata is kept for as long as required for invoicing and Polish tax and accounting obligations, up to 5 years after the end of the year in which the invoice was issued, and is then deleted.
4. Where processing happens
Requests pass through our API gateway and are processed on dedicated GPU capacity that we rent from cloud infrastructure providers. The current list of infrastructure providers is in the Privacy Policy. The datacenter countries for each model are published in our /v1/models endpoint.
5. Security
All traffic is encrypted in transit with TLS. Access to the API requires a secret API key. Access to our servers is limited to the business owner.
6. Changes
We will announce material changes to this policy at least 30 days in advance to customers with active API keys.
7. Contact
MADD Marcin Kisielinski, ul. Kajki 10-12, 10-547 Olsztyn, Poland. VAT ID (NIP) PL7422297084. E-mail: domaai@madd.im.