Skip to main content
NagaAI does not use your prompts or outputs for training. It stores request and response bodies only when request logging is on for the workspace, for 30 days; request metadata is kept for one year. The legal terms are in the Privacy Policy.

Request logging

Each workspace decides whether its requests keep their bodies, on its Observability page in the dashboard, as on OpenRouter. With Request logging on, NagaAI stores the full request and response bodies of the workspace’s requests, from its API keys and the Playground. Logs shows them. It’s off for a new workspace. The workspace’s admins turn it on or off, and the change applies within seconds to requests sent afterward. Bodies already stored stay until their 30 days run out.

Data NagaAI keeps

NagaAI records metadata for every request. Request and response bodies appear in Logs only for workspaces with request logging on. Security and operational logs, such as the IP addresses used for rate limiting and error diagnostics, are covered by the Privacy Policy.

Model providers

Each request is processed by a third-party inference provider under its own terms. NagaAI cannot guarantee that the provider does not retain the request or use it for training. If you need that guarantee, do not send data through NagaAI that may not be shared with third parties. This holds for every model, free or paid. You can’t see which provider served a request, and it can change between requests, as Routing explains.