guardyn.dev
Pricing

Pay for what the models actually cost, plus a margin we publish.

One credit is one US cent of model value. Your plan includes an allowance; heavier months are a top-up, not a renegotiation. Endpoints and seats are counted, not metered.

Free

One machine, real answers

$0/month

Enough to run a genuine compromise assessment on a machine you own.

  • 2 endpoints
  • 200 credits a month
  • Fast models
  • Full read-only tool set
  • 14-day history
Start free

Starter

For the person who owns the estate

$49/month
or $470/year — two months free

A small fleet, the balanced models, and history that outlives the incident.

  • 15 endpoints
  • 5,000 credits a month
  • Balanced models (Sonnet 5, GPT-5.6 Terra)
  • 2 concurrent investigations
  • 90-day history
  • 2 API keys
Choose Starter

Team

For a security function

popular
$199/month
or $1910/year — two months free

Frontier models, real concurrency, exportable audit trail.

  • 100 endpoints
  • 25,000 credits a month
  • Frontier models (Opus 5, GPT-5.6 Sol, Gemini 3.1 Pro)
  • 5 concurrent investigations
  • Scheduled sweeps
  • Audit-log export
  • 180-day history
Choose Team

Enterprise

For the whole estate

Talk to us

Every model including the security-specialist tier, SSO, your own provider keys.

  • Unlimited endpoints
  • Specialist models (GPT-5.6 Cyber, Fable 5)
  • Bring your own provider keys
  • SSO and SCIM
  • 20 concurrent investigations
  • 2-year history
  • Priority support
Contact sales
What a credit buys

The arithmetic, in the open.

A credit is a cent of retail model value. Each model carries its own price and its own margin, both visible to whoever administers the platform.

ModelProviderTierInput / MtokOutput / MtokNeeds

A typical single-question investigation on the balanced tier — four or five tool calls and a couple of pages of answer — lands between 8 and 25 credits. A full compromise assessment on the frontier tier is usually 60 to 200. Prompt caching means follow-up questions in the same conversation cost a fraction of the first.

Questions

The things people ask before signing up.

What happens when I run out of credits?

Investigations stop with a message saying how many credits the next one needs. Nothing is deleted, nothing is throttled silently, and your agents stay connected. Top up or wait for the next period.

Do you train on my data?

No. Evidence from your machines is sent to the model provider you selected in order to answer your question, and nowhere else. Nothing is used to train anything.

Can the agent be used to attack my machines?

It has no tool that executes anything and no tool that writes anything. The tool set is compiled into the binary — this server can name a tool, never supply a command. Worst case, a compromised server can read from your fleet.

Can I use my own provider keys?

On Enterprise, yes: your Anthropic, OpenAI or Google key, your provider bill, and you pay us for the platform rather than for tokens.

Is there a minimum term?

No. Monthly plans cancel at the end of the period and keep working until then.

Where does it run?

guardyn.dev is a hosted service. You install the agent on your machines; everything else runs on our infrastructure, so there is nothing for you to operate, patch or back up. The agent is read-only and outbound-only, which is what makes that acceptable — see the security model.