Skip to content
Cloud Claude

Priced on compute.

One unit. Three tiers. Starts at $2,500 per month. Sandboxes, managed Postgres, agent runtime, and background workers all draw from the same pool.

Join the waitlistTalk to us

Overage rates are published, not hidden. Same rates on every tier.

The unit

Everything runs on compute-hours.

No per-resource metering table. No line-item bill. One number to watch, one dial to tune.

Formula

1 compute-hour = 1 vCPU × 1 GB RAM × 1 hour

Your monthly pool covers everything: agent runtime, Postgres, background jobs, sandboxes. When you exceed it, overage bills at the published rate.

Plans

Three tiers. Substantial by design.

No free tier. No pay-per-seat. Priced against the compute your workloads actually use.

  • Team

    $2,500/mo

    For engineering teams putting their first agent workloads in production.

    • 3,000 compute-hours per month
    • 10 team seats, roles and audit log
    • 500 GB storage, 5 TB egress
    • Per-project isolated networks and managed Postgres
    • Standard uptime SLA, email support next business day
    • Same platform Enterprise runs on.
    Join the waitlist
  • Recommended

    Business

    $12,000/mo

    For teams running production agents across staging and prod, with SSO and audit.

    • 15,000 compute-hours per month
    • 50 team seats
    • 2 TB storage, 25 TB egress
    • SSO / SAML included
    • Staging and production environments isolated by default
    • Shared Slack channel, 4-hour response
    • Enhanced uptime SLA, audit retention 1 year
    Join the waitlist

  • Enterprise

    From$50,000/mo

    For platform teams at scale. Annual commitment, custom terms.

    • Committed compute pool with volume pricing
    • Unlimited seats
    • Dedicated Firecracker fleet on pinned hardware
    • Financial-credit SLA
    • Named CSM, solutions engineer, 1-hour response
    • SOC 2 readiness, HIPAA and PCI reviews on request
    • Private cloud and BYOC options
    Talk to sales

Beyond the pool

Overage rates, published up front.

Same rates on every tier. Alerts and hard caps live in the console.

  • Compute overage
    Draw beyond your included pool at the published rate.
    $0.12 / compute-hour
  • Egress overage
    Outbound bandwidth past the tier allowance.
    $0.05 / GB
  • Storage overage
    Managed Postgres and volume storage past allowance.
    $0.15 / GB · month
  • Dedicated Firecracker fleet
    Single-tenant microVMs on pinned hardware.
    Custom

Included on every tier

You’re not paying for the platform. You’re paying for the compute.

Every tier — from Team to Enterprise — runs on the same production platform. What changes across tiers is scale, support, and commitment.

  • HTTPS and custom domains

    Managed TLS on every custom domain, no ACME to run.

  • MCP for agents

    90+ MCP tools, per-role capabilities, full audit trail.

  • Isolation by default

    Firecracker microVMs, per-project networks, per-tenant Postgres.

  • Managed Postgres

    Automated image backups, one-click rollback, terminal access.

Enterprise

Building on Cloud Claude at scale?

Enterprise plans start at $50,000 per month on annual commitment and scale into eight figures. Talk to our team about dedicated infrastructure, compliance workflows, and volume compute pricing.

Committed compute pools, custom SLA, dedicated Firecracker fleet, named CSM. Priced against the workloads you actually run.

Contact sales

Common questions

Answers to the questions procurement asks first.

Why one compute-hour unit instead of per-resource billing?
Because you shouldn’t need a spreadsheet to understand your bill. One number rolls up sandboxes, managed Postgres, agent runtime, and background workers — the same way your platform team plans capacity internally.
Is there a free tier?
No. Cloud Claude is built for teams and enterprises running production workloads. If you’re evaluating for a specific use case, contact sales for a scoped proof-of-concept.
How do compute-hours map to real workloads?
One compute-hour equals one vCPU × one GB of RAM × one hour. A typical Node service on 1 vCPU and 512 MB running 24/7 consumes about 360 compute-hours per month. Larger agents, batch jobs, and databases consume proportionally from the same pool.
Can we commit annually for a discount?
Yes. Team and Business tiers offer 15% off on annual commitment. Enterprise is annual by default and priced against your committed compute pool.
What happens if we exceed our pool?
Your workloads keep running. Overage bills at the published rate above. Set usage alerts and hard caps from the console so surprises never reach your inbox.
Do you offer air-gapped or on-premise deployments?
Yes, under Enterprise. Talk to sales about dedicated regions, FedRAMP considerations, and BYOC on AWS or GCP.

Priced for teams that ship.

Join the waitlist and we onboard you in small cohorts, starting with the customers running the largest compute pools.

Join the waitlistTalk to us