• Playground
  • Models
  • Rankings
    • About Boundless
    • Quickstart
    • Coding Agents & Harnesses
    • API Compatibility
    • System One
    • MPP & Agentic Payments
    • Credits & Billing
    • Pricing
    • Rate Limits
    • Teams
    • Privacy & ZDR
    • FAQ
  • Support
All systems operational
© 2026 BoundlessTermssupport@boundless.network

Privacy & ZDR

Zero data retention: we never store your prompts or outputs, never inspect them, and never train on them. Only billing metadata is kept.

Your data, your rules. Privacy is built into how Boundless works: requests are processed in memory and then discarded. For the full legal wording, read Section 3.2 of the terms.

What we never do#

  • Store your tokens. Prompts, outputs, and request or response bodies are never written to a log, a database, or any persistent storage.
  • Inspect your prompts. No person reviews what you send, and nothing is sampled for analysis.
  • Train on your data. Your inputs and outputs are never used to train or tune any model.

What we keep for billing#

To charge you exactly, we keep a little metadata about each request: token counts, model, timestamp, cost, and which key, team and user made the request. None of it contains your content. The per-request record is deleted after 90 days, and you can opt out of it entirely by contacting support.

The usage page shows the same numbers, so you can recompute any charge yourself.

Exceptions#

Frontier closed models, such as those from Anthropic and OpenAI, cannot be ZDR. That is a limit of their upstream providers, not of Boundless: those providers may retain prompts and outputs under their own policies. Our side of every request stays ZDR, but do not use these models when zero data retention is required end to end.

The legal commitment#

This page is the plain-language version. The binding text is Section 3.2 of the terms, Zero Data Retention.

NextFAQ