Privacy & Data Handling

Your data stays private

CheapestInference is designed with privacy as a default. Here’s how we handle your data:

Prompts and completions

• Never stored to disk. Your prompts and model outputs are processed in memory during inference and discarded immediately after.
• Never used for training. We do not train any model on your data. Ever.
• Never shared with third parties. We exclusively serve open-source models. Your data never leaves our inference infrastructure to reach third-party model providers.

What we do store

• Account data — email, subscription status, API key metadata (name, plan, rate limits).
• Usage metadata — request counts, token counts, and cost per key. We do not store the content of requests or responses.
• Payment information — processed by Stripe (card payments) or verified on-chain (USDC on Base). We do not store card numbers.

What we log

• Request metadata only — timestamps, model used, token counts, response status codes.
• We do not log prompt content or model outputs under normal operation. In rare cases, a small portion of requests may be inspected for debugging or security purposes.

Compliance

• GDPR compliant. EU users can request data export or deletion.
• SOC 2 Type II and ISO 27001 certified infrastructure.
• Data processed in secure, audited datacenters.

Data retention

• Account data — retained while your account is active. Deleted within 30 days of account deletion.
• Usage metadata — retained for billing and analytics. Aggregated and anonymized after 90 days.
• Inference data (prompts/completions) — not retained. Processed in memory only.

Per-key privacy

If you create API keys for your users via the Management API, each key’s usage is tracked independently. You can see per-key usage (token counts, costs) but you cannot access the content of your users’ requests or responses.

Contact

For privacy-related questions or data requests, contact [email protected].