Pricing
Open-core, honestly priced.
The full AIFenders engine is free and open source — five rails, declarative policy, zero runtime dependencies, self-host anywhere. When you want it hosted, our multi-tenant Cloud is usage-based. When you need a VPC deployment with SSO and an SLA, Enterprise fits around your controls. Same engine underneath, every tier.
Open Source
Free, self-hosted
The full engine and all five rails. Pure Node.js, zero runtime dependencies — trivially self-hostable in your own environment. Community support.
Read the docs- ✓ All five rails — input, dialog, retrieval, execution, output
- ✓ Self-host anywhere — VM, container or VPC, no phone-home
- ✓ Bring your own model — any OpenAI-compatible provider or local Ollama
- ✓ No telemetry — declarative JSON policy, your data never leaves
Cloud
Usage-based
Hosted multi-tenant AIFenders. Get an aif_live_ key in minutes, swap one base_url, and let us run upgrades and scaling for you.
Start freeStart free, pay as you scale.
- ✓ Instant API key — aif_live_ credentials, live in minutes
- ✓ Managed upgrades — hosted, patched and scaled for you
- ✓ Dashboard & console — policies, audit log and benchmarks
- ✓ Multiple policies — per-key named policies, BYO providers
Enterprise
Custom
Self-host or run in your own VPC for data residency, with the governance and support controls a regulated buyer requires.
Contact sales- ✓ Self-host / VPC — data stays in your region and controls
- ✓ SSO / SAML + RBAC — identity governance for your org
- ✓ Compliance packs — audit export and auditor-ready mapping
- ✓ Dedicated support — SLA and deployment assistance
Compare
What's in each tier.
The engine is identical everywhere — the difference is who runs it and which governance controls come with it.
| Capability | Open Source | Cloud | Enterprise |
|---|---|---|---|
| All five rails | ✓ | ✓ | ✓ |
| Self-host | ✓ | — | ✓ |
| Hosted cloud | — | ✓ | — |
| Bring your own provider | ✓ | ✓ | ✓ |
| SSO / SAML | — | — | ✓ |
| Compliance packs | — | — | ✓ |
| SLA | — | — | ✓ |
FAQ
The questions security teams ask first.
Is my data used for training?
No. AIFenders never trains on your prompts, completions or policy data. Raw secrets are never echoed into responses, telemetry or the audit log.
Can I use my own model and keys?
Yes — bring your own provider and key per policy. Any OpenAI-compatible provider, Anthropic, or a local Ollama works. Keys are masked to the last four and never returned.
Do you see my prompts?
Self-host keeps prompts entirely in your own environment. In gateway mode AIFenders holds no external provider key, and PII is masked before the model and before the user.
What does "usage-based" mean?
Cloud is priced on the requests and volume you send through it — no seat minimums. You start free and pay as your traffic scales.
Start free. Scale when you're ready.
Self-host the open-source engine today, or launch the console and get an aif_live_ key in minutes. Same five rails, one base_url swap.