Pricing.
Built once. Owned forever. No metering. No per-token bills. No surprise invoices.
Built once. Owned forever. No metering. No per-token bills. No surprise invoices.
No subscription on the hardware. The appliance keeps running with or without us.
More memory and bandwidth for larger contexts and heavier workloads. No subscription on the hardware.
Five years of security patches ship with the appliance. The licence is for ongoing upgrades: new models, new agents, new Syntex OS features. Cancel any time; the appliance keeps running. No remote kill-switch. Volume licensing available on request for 5+ units.
Built to order: each appliance is provisioned, benchmarked and validated, then ships within 10 business days of order confirmation. Founding-batch launch pricing while the first run lasts.
£1,200 / year is the per-appliance list price. Volume licensing for 5+ units is custom; we’ll quote when confirming your reservation.
Thank you. We’ll confirm your appliance and next steps by email shortly.
If our email doesn’t reach your inbox within a few minutes, please check your spam or junk folder; arsenale.ai is a new sender and some mail providers route first-time messages there. Marking it as “Not Spam” helps future messages arrive cleanly.
We couldn’t submit your reservation just now. Please try again, or email contact@arsenale.ai directly with your configuration.
Comparable sovereign on-premise inference: a 96GB Apple Silicon machine starts at £4,199, hardware only, no model, no runtime, no agents, no support. An NVIDIA H100 deployment starts at £35,000+, GPU only. Both ship as raw compute. Months of engineering still stand between them and a working autonomous business. Arsenale ships complete: hardware, model, runtime, agents, learning, support; £2,499.
Custom deployments, air-gapped configurations, C2/ISR workflows, volume licensing.
contact@arsenale.ai →