Managed AI Hosting PaaS

Agent Starter

$19.00 /mo
Order Now
  • BYOK (bring your own key, zero markup)
  • Hardened container isolation, 3-day log retention
  • Community / email support

Over-allowance behaviour: warn 80/100% -> throttle -> block. No "unlimited", no surprise bills. Metered overage / prepaid credits available.

Managed AI Hosting on AKT Agent Cloud - for solo builders shipping one agent or app.

  • Burst up to 0.5 vCPU / 1 GB RAM
  • 75 vCPU-hr included compute / mo
  • 5 GB persistent storage (EFS + S3)
  • 40 GB egress / mo, metered beyond
  • 1 concurrent agent, 1 concurrent job, 2 scheduled jobs, 1 worker
  • Rate limit 10 req/s
  • Scale-to-zero - idle costs nothing
  • AI

Agent Business

$49.00 /mo
Order Now
  • BYOK + optional managed AI gateway (small included tokens + metered)
  • Hardened container isolation, 7-day log retention
  • Email support

Over-allowance behaviour: warn 80/100% -> throttle -> block. No "unlimited", no surprise bills. Metered overage / prepaid credits available.

Managed AI Hosting on AKT Agent Cloud - for teams running multiple agents with room to grow.

  • Burst up to 1 vCPU / 2 GB RAM
  • 350 vCPU-hr included compute / mo
  • 15 GB persistent storage (EFS + S3)
  • 120 GB egress / mo, metered beyond
  • 3 concurrent agents, 3 concurrent jobs, 10 scheduled jobs, 3 workers
  • Rate limit 50 req/s
  • 1 always-warm service + scale-to-zero for the rest
  • AI

Agent Pro

$99.00 /mo
Order Now
  • BYOK + managed AI gateway (medium included tokens + metered credits)
  • Hardened container isolation, 14-day log retention
  • Priority email support

Beyond the included allowance = pay-as-you-go metered overage or prepaid credits. No "unlimited", never a silent bill.

Managed AI Hosting on AKT Agent Cloud - for serious agent workloads with metered headroom.

  • Burst up to 2 vCPU / 4 GB RAM
  • 750 vCPU-hr included compute / mo + metered overage
  • 40 GB persistent storage (EFS + S3)
  • 250 GB egress / mo + metered overage
  • 6 concurrent agents, 8 concurrent jobs, 30 scheduled jobs, 6 workers
  • Rate limit 200 req/s
  • 2 always-warm services + scale-to-zero for the rest
  • AI