SOC 2-ALIGNED  ·  EU AI ACT  ·  NIST AI RMF  ·  ISO 42001-READY [email protected]
Odingard / Cerberus Book a Demo
Pricing

Priced per production environment.

One annual license covers one production deployment — every agent, model and provider behind it, and as many gateway replicas as you need to run it safely. Not per seat, not per model, not per call.

Free · MIT

Open Core

$0

The detection engine, free forever. Wrap your executor with guard() or run the HTTP gateway. LangChain, Vercel AI SDK, and OpenAI Agents adapters included.

  • ✓ L1–L3 runtime detection
  • guard() SDK + HTTP gateway
  • ✓ Framework adapters
  • ✓ Community support
npm install @cerberus-ai/core
Annual license

Standard

$12,000/year

1 production environment

Everything in Open Core, plus the durable ledger, containment, and audit surfaces — all running inside your own VPC. Data never leaves your infrastructure.

  • ✓ Durable SQLite provenance ledger (L4)
  • ✓ Blast-radius containment & quarantine
  • ✓ Egress / SSRF guard, fail-closed gateway auth
  • ✓ Signed, audit-ready evidence export
  • ✓ Grafana + OpenTelemetry stack included
  • ✓ Unlimited agents, models and replicas
  • ✓ Onboarding & setup support
Annual license

Premium

$36,000/year

Up to 3 production environments

For teams running Cerberus in more than one production environment — separate regions, business units, or a segregated regulated workload.

  • ✓ Everything in Standard
  • ✓ Up to 3 production environments
  • ✓ Priority support
  • ✓ Guided onboarding & deployment review
Custom

Talk to us

Quote

4+ environments or air-gapped

Larger fleets, air-gapped deployments, custom SLAs, security questionnaires and MSAs.

  • ✓ 4+ production environments
  • ✓ Air-gapped / offline licensing
  • ✓ Custom SLA and support terms
  • ✓ Procurement, MSA and security review
Contact Sales ↗

Prices in USD, billed annually. Setup support is included with every license. Marketplace procurement via Google Cloud & Azure is on the roadmap. Questions before you buy? [email protected] — we reply within one business day.

FAQ

Common questions.

What counts as a production environment?

One deployment of the Cerberus gateway serving production traffic — typically one VPC, one region, one business unit. Run as many replicas behind your load balancer as you need for resilience; that is one environment, not several. Staging and development deployments are included at no charge.

Is it priced per model, per agent or per seat?

None of those. One gateway protects however many agents, models and providers you point at it — OpenAI, Anthropic, Google, local models, or all of them at once. You should not pay more for putting Cerberus in front of more of your estate.

Why annual and not monthly?

A lapsed license takes the gateway out of your production path. Annual billing means that decision happens once a year with notice, not on a card expiry. Evaluate for free first — the MIT core has no time limit.

Can I try it before buying?

Yes. The MIT-licensed core is free: npm install @cerberus-ai/core. Run the live demo and the in-repo proof path with no license required.

Where does my data go?

Nowhere. The Enterprise gateway is self-hosted in your own VPC. Cerberus never sees your traffic or data.

What happens after I pay?

Your license key is emailed immediately. It authenticates the release download, the container image pull and the running gateway — no GitHub account or registry token needed.

What does onboarding look like?

Every license includes hands-on setup with our engineering team — from image pull to first blocked attack in your environment.