Pricing
Protect endpoints with Beacon, or build and run agent workflows with Platform.
USD monthly rates, excluding applicable taxes.
Monthly plans
Beacon
$6
per device / month
$30 monthly minimum covering up to five devices.
Monitor and protect agents on your endpoints. No Platform purchase required.
- Agent, tool and MCP inventory, coverage and Exposure Report
- Supported policies and detections, fleet console and audit access
- 30-day audit history within the included capacity
- No organizational knowledge store or hosted execution included
- Documentation and email support
Platform
$299
per organization / month
Five Beacon devices and all human collaborators included.
Build and run connected agent workflows across your business.
- Beacon plus workflows, shared knowledge, policy and credential controls
- Unlimited customer-managed runners. Connect your cloud or infrastructure.
- SAML/OIDC single sign-on and SCIM provisioning included.
- 10 GB of knowledge storage
- 30-day audit history within the included capacity
- Alyria-hosted runners: 1,000 active vCPU-minutes/month and two concurrent standard runners
- Documentation and email support
Enterprise
Scoped
to your requirements
We’re security practitioners too. We hate paying extra for SSO and identity lifecycle management, so we include them. No enterprise upgrade required.
Enterprise pricing reflects your usage, deployment and support requirements, with discounts for term agreements. It is not a paywall for security features.
- Agreed deployment, residency or integration requirements
- Agreed capacity, audit retention and support commitments
Usage rates
Connect your own cloud and model accounts, or opt into limited Alyria-hosted runners and model usage. Managed usage and extra capacity require opt-in.
| Usage | Rate | Scope |
|---|---|---|
| Additional devices | $6/device/month | Beyond the five devices covered by Beacon's minimum or included in Platform. Platform's included devices are not billed again. |
| Optional Alyria-hosted execution | $0.04/active vCPU-minute | After Platform's 1,000 included minutes/month. Standard runner: one vCPU, up to 2 GB memory; two concurrent runs included. |
| Optional model usage through Alyria | Provider rate + 25% | Includes optional model-backed semantic evaluation. Provider input, output and cache rates apply; applicable cache discounts pass through. |
| Additional retained knowledge or audit capacity | $2/GB-month | Above the allowance for each separate pool, measured by daily-average retained logical storage. More capacity does not extend audit retention. |
Reports and assessments
These are separate from the software plans.
- Agent Exposure ReportFree
- See the agents and tools running on your endpoints. Get a free Agent Exposure Report
- Team and use-case assessmentsFree
- Answer the questions for your function and see your result on the page. Explore free team assessments
Questions
- Can we use our own cloud and model accounts?
- Yes. Customer-managed runners use your own cloud and model accounts, and those providers bill you directly. Platform includes unlimited customer-managed runners, with no Alyria per-runner license or hosted-compute charge for your own compute; the Platform subscription still applies. Model accounts, runtime and customer-managed encryption keys are separate choices. Holding an encryption key does not supply model credits or compute.
- What counts as managed execution?
- Allocated execution time, including network or model waits while a runner is held. Queue, approval and suspended time do not count when compute is released. A standard Alyria-hosted runner has one vCPU and up to 2 GB memory. Larger-memory, GPU, dedicated and persistent-idle runners require separate scope.
- How much audit history is included?
- Both paid plans include 30 days of audit history within 1 GB per organization plus 0.25 GB per paid-plan device. This covers audit metadata and evidence references, not unlimited prompts, attachments or recordings.
- How are devices and collaborators counted?
- Devices are counted by enrolled identity, not agent count or online status. Platform includes all human collaborators and five device licenses; installing all five is not required. There is no developer-seat meter or fee for each raw tool call. Model inference is separate from device protection, and import or embedding model use is separate from storage.