Privacy Policy

1. Scope and Overview

This Privacy Policy applies to the AI inference services, APIs, and web platforms operated by Infergator AI ("Infergator", "we", "us", or "our"), including endpoints hosted at api.infergator.com and related domains. By utilizing our inference endpoints directly or via third-party orchestrators (e.g., OpenRouter), you agree to the data handling practices described in this policy.

2. Processing of Customer Prompts and Model Outputs

When you dispatch requests to our inference API:

• Ephemeral In-Memory Lifecycle: Prompt tokens are received over an encrypted TLS connection directly into volatile accelerator memory (GPU VRAM / Host RAM). Tokens exist solely for the duration of the compute forward-pass required to generate the output stream.
• Instant Eviction: Once generated tokens are streamed back via Server-Sent Events (SSE) or HTTP chunked transfer to the client, all intermediate tensor caches, key-value (KV) states, and token buffers are evicted and wiped from memory.
• Zero Persistent Storage: No prompt data, system prompts, chat history, or completions are written to non-volatile disk drives, relational databases, log aggregators, or telemetry queues.
• No Model Training or Fine-Tuning: Infergator AI strictly never uses customer data, prompts, or completions to train, retrain, distill, or evaluate any artificial intelligence models.

3. Telemetry and Operational Metadata

To ensure service reliability, rate limiting, and accurate token billing reconciliation with upstream orchestrators (such as OpenRouter), we maintain minimal, non-content operational telemetry:

• Numeric prompt token count and completion token count.
• Request timestamp, HTTP status code, and latency metrics (Time to First Token, tokens per second).
• Model identifier requested (e.g., deepseek-v4.1-flash-xolotl).
• Pseudonymous routing identifiers provided by upstream gateways for billing verification.

This operational metadata contains no textual prompt content, personally identifiable information, or completion outputs.

4. Network Security & Encryption

All network traffic to and from Infergator AI is encrypted in transit using industry-standard Transport Layer Security (TLS 1.3). Edge transport is hardened through Cloudflare Anycast infrastructure with automated DDoS mitigation, origin authentication, and strict QUIC/HTTP3 protocol enforcement.

5. Third-Party Marketplace Integration (OpenRouter)

When you access Infergator AI through OpenRouter, OpenRouter acts as an intermediary router and billing provider. OpenRouter's privacy practices govern their handling of your account credentials and routing preferences. Infergator AI receives requests routed from OpenRouter's edge under OpenRouter's Zero Data Retention provider tier.

6. Contact Information and Data Protection

If you have questions, security inquiries, or require formal data protection agreements (DPA), please contact our security and engineering team at: