InferStack

Privacy Policy

Effective 1 November 2026

InferStack ("we") provides AI model inference through an API, mainly to customers of OpenRouter. This policy explains what we process, what we keep, and for how long.

1. Prompts and completions

2. What we do keep

For billing, capacity planning and abuse prevention we keep request metadata:

DataRetention
Request ID, timestamp, model, input/output token counts, latency, HTTP status, amount billed13 months
Aggregated metrics (request rates, error rates, throughput), with no per-request content30 days

Request metadata does not contain prompt text, completion text or images.

3. Where processing happens

4. Who we share data with

5. Your rights

Because we do not store request content, there is generally no personal content for us to access, correct or delete. For questions about request metadata, or a complaint under Canada's Personal Information Protection and Electronic Documents Act (PIPEDA), contact hello@inferstack.ca.

6. Security

All traffic is encrypted in transit (TLS). GPU servers accept no inbound connections from the internet and are reachable only over a private encrypted network from our gateway.

7. Changes

We will post updates on this page with a new effective date.