Short version: requests are processed in memory and are not logged, stored, or used to train models.
We do not use customer prompts or completions to train, fine-tune, or evaluate models. No human reviews request content except where legally compelled.
Payload retention: zero. Usage metadata: 90 days. Billing records: as required by tax law.
Inference runs on Nebius AI Cloud GPU instances. Billing for OpenRouter-distributed traffic is settled by OpenRouter under its own terms. Email is delivered through Oracle Cloud Email Delivery.
Automated abuse classification may inspect request metadata. Content is retained only if flagged for a legal or safety report, and only for as long as the report requires.
Write to admin@inference.chessurge.org for access, deletion, or a data processing agreement. Enterprise customers can request zero-retention attestation.