Inference

1. Data Collection & Telemetry

At Inference, we believe in radical transparency. We only collect the minimal amount of telemetry data required to provide you with our intelligent routing services, monitor system health, and calculate billing metrics. This includes your account information (email, name), API key identifiers, and standard request metadata (IP addresses, request sizes, latency metrics, and aggregate token counts).

2. Prompt Data & Privacy

Your prompts are your property. We route your prompts to the requested underlying models via encrypted tunnels, but we do not train on your prompt data. We have strict data residency and caching rules based on your selected routing tier:

Premium Tier: Prompts routed through our Premium tier are strictly ephemeral. They are processed in memory and are never persisted to our databases.
Standard Tier: Prompts routed through our Standard tier may be temporarily cached for up to 24 hours strictly to improve global latency and system performance.
We never sell, rent, or lease your prompt data or user information to third parties under any circumstances.

3. Third-Party Integrations

By using Inference, your prompt data is necessarily transmitted to the underlying model providers (e.g., OpenAI, Anthropic, Meta) based on your dynamic routing configurations. Those specific providers have their own data retention and privacy policies which you are subsequently subject to. We enforce that all outgoing connections to these providers are encrypted via TLS 1.3 or higher.

4. Contact & Data Deletion Rights

Under GDPR and CCPA regulations, you have the right to request a complete deletion of your account and all associated telemetry data at any time. You can initiate this automated "Right to be Forgotten" process directly from the "Danger Zone" in your dashboard Profile settings, or by contacting our data protection officer at [email protected].