Simple, transparent pricing
Pay only for what you use. No hidden fees, no object taxes. Start with $100 in free credits.
No objects tax. Clusters and services are completely
free. You only pay for CPU, RAM, and ephemeral storage usage.
Credit-based billingCost Calculator
Monthly Cost Breakdown
CPU (1 vCPU × 2,592,000s) $259.20
Memory (2GB × 2,592,000s) $259.20
Storage (10GB × 2,592,000s) $259.20
Total per month $777.60
Free Credits
Start with $100 in free credits. This configuration would last approximately 0 months with your free credits.
Usage pricing
Compute
- • vCPU-seconds (1-minute minimum per container run)
- • RAM-seconds (1-minute minimum per container run)
Storage
- • Ephemeral storage GiB-hours (per-second metering, 1-minute minimum)
$0 for these
- • Clusters, Services, Jobs, CronJobs as API objects
Credits
- • Sign-up credit applied at activation
- • Auto top-ups (configurable) once a card is on file
- • If credits reach $0 with no card: running workloads are terminated by design
Credit-based billing
Free credits
Start with $100 in free credits when you sign up. No credit card required to begin.
Automatic top-up
Set up automatic credit top-up rules after adding a credit card to avoid service interruption.
Safety by design
When credits are exhausted, all workloads are automatically terminated. This prevents unexpected charges and ensures you stay within your budget.
FAQ
Stateless web/API Services, Jobs, and CronJobs. Event-driven tasks, scheduled batch, dev/test/demo environments, and checkpointable compute. State lives in your own DB/object storage.
Not at launch. Local disk is shared and optimized for ephemeral data. Use external databases and object stores for persistence.
Workloads are private by default. Expose them via the managed global HTTP load balancer using simple path-prefix rules. For UDP-only apps, publish as a DNS-type Service (multi-A DNS); for HTTP apps, use load-balanced Services with health checks and zero-downtime deploys.
A Cluster defines geography and isolation. You choose the continents/regions for each Cluster (e.g., dev in EU, prod in US+EU). Services live inside a Cluster and inherit its placement and guardrails.
If there's no payment method on file, workloads stop. After you add a card, you can enable auto top-ups to avoid interruptions.
Metrics are retained with sensible windows (e.g., short-window for billing accuracy, per-minute for dashboards). Logs stream directly from nodes on demand for fast troubleshooting, without the cost of always-on log shipping.
Yes—optionally run with VM isolation (Kata/QEMU) for stronger sandboxing. Default networking is deny-by-default, with tenant-scoped DNS, WireGuard-encapsulated pod traffic, and encrypted platform volumes (LUKS2).
Occasionally, yes. You'll receive a termination signal and a replacement replica is scheduled per policy. Architect for horizontal scaling and graceful shutdown.
Per-container bandwidth limits apply, and there's no IO throttling yet on shared local storage. If you need sustained high bandwidth or strict disk QoS, this isn't the right fit—yet.
Default public domains are provided. Custom domain automation is on the roadmap (ACME). For now, manual DNS challenges may be supported where needed.
You get one monthly invoice per account, plus per-cluster breakdowns so teams can see what costs what.
A compliant, managed secrets/KMS path is planned; initially use your external KMS or secret manager.