



Choose a plan (monthly + tax until you cancel; charged only when you confirm at checkout):
Shared, burstable vCPUs up to your plan’s count (our workloads have priority)
RAM: up to your plan’s limit (hard limit, no swap). Host RAM is oversubscribed and not set aside for you: when it runs short, processes may be stopped and your node restarted before the limit; in-memory data is lost; disk files stay
Backups: On-site hourly snapshots, kept up to 14 days; restores on request, best-effort, and not a substitute for your own backups.
Root in your own isolated gVisor container; no access to the host or other customers; runs on CPU
Hermes Agent preinstalled: 100M-400M tokens/mo by plan, no rollover. At the limit, requests are refused until the monthly reset
Shared and preemptible: you may get less than your plan’s vCPU or upload rate; pro-rated credit for sustained preemption or extended outages, and for host memory shortages that block use over 24 h in a billing period; no uptime SLA
Shared upload; 250 GB/mo egress cap; no hosting, seedboxes or mining
US customers only. Sold by SlowGlow AI LLC (NC). Terms, AUP and Privacy Notice apply.