Start with one team and a browser extension; grow to every surface — browser, desktop, and your own AI apps — under one policy. Simple per-user pricing, no per-prompt metering, no surprises.
For a team getting started with AI, safely.
For security teams governing AI across the org.
For regulated orgs and large, self-hosted deployments.
The same engine ships as the GPT-Shield SDK and a drop-in OpenAI / Anthropic-compatible gateway. Wrap a client in one line and every call to your internal models is protected — pseudonymized on the way in, restored in the reply, with full lineage. Included from Business with a generous monthly request volume; Enterprise adds self-hosting and custom volume — priced on protected requests, never per token.
Talk to us about internal AI →| Feature | Team | Business | Enterprise |
|---|---|---|---|
| Browser extension + desktop agent | ✓ | ✓ | ✓ |
| Data-Shield + File-Shield | ✓ | ✓ | ✓ |
| Internal AI protection — SDK + gateway | – | ✓ | ✓ |
| Input- + Response-Shield | – | ✓ | ✓ |
| Trace-Shield dashboards | – | ✓ | ✓ |
| Central policy management | – | ✓ | ✓ |
| SSO (Google / Microsoft) | – | ✓ | ✓ |
| SAML SSO + SCIM | – | – | ✓ |
| Self-hosted deployment | – | – | ✓ |
| SOC 2 report + DPA | – | – | ✓ |
| Support | Priority | Dedicated + SLA |
Yes — Team and Business include a free trial so you can roll it out to a team and see exactly what crosses the boundary before you pay.
An active person protected by GPT-Shield in a billing month — someone with the extension, desktop agent, or gateway in front of their AI usage. Inactive accounts aren't billed.
No. Pricing is a flat per-user rate. Detection runs locally, so there's no per-prompt cost to pass on — protect as much as you want.
Per-user pricing covers the people you protect on ChatGPT, Claude, Gemini and the like. Protecting your own internal apps and agents runs through the SDK and gateway, included from Business with a generous monthly request volume and priced on protected requests beyond that — never per token. Enterprise sets custom volume and self-hosting. Tell us your call volume and we'll size it.
Yes, on Enterprise. The gateway is self-hostable and OpenAI/Anthropic-compatible, and the engine runs entirely in your environment.
Anytime. Upgrade as you expand coverage across surfaces and teams; changes are prorated.
We do — reach out at noah@gpt-shield.com and we'll sort you out.
Book a 20-minute walkthrough, or grab an early-access invite — pricing is easy once you've seen it work.
No raw data leaves your machine. Ever.