new · L4 and A100 GPU pools live in blr1 and fra1new · cost-guard agent now proposes committed-use plans4,912 workloads migrated off AWS this quarter99.99% SLA on every tier, Hobby includedDNS p50 resolve 38ms across 31 PoPs
02 / Compute engine

Managed like a panel. Priced like a VM.

Cloudways-grade managed stacks, except the OS, runtime, TLS, backups, and scaling policy are agent-owned. You get root if you want it, and never need it.

n4-micro
$6
1 vCPU · 1 GB · 25 GB SSD
n4-small
$18
2 vCPU · 4 GB · 80 GB SSD
n4-std
$42
4 vCPU · 8 GB · 160 GB SSD
n4-gpu-l4
$0.62/hr
L4 24 GB · 8 vCPU · 32 GB
// n4n.toml — the whole config
[app]
name = "checkout"
size = "n4-std"
[scale]
min = 2 max = 8
target_cpu = 65
[agents]
right_size = "propose"
failover = "auto"

What you don't write

Dockerfiles — buildpacks detect node, python, php, go, rust, laravel, rails
nginx configs, TLS, HTTP/3, or cert renewal
Backup cron, restore runbooks, or staging clones
Kernel patch windows or OS upgrades
Load-balancer health checks or blue/green scripts
Managed Postgres, Redis, and MySQL attach in one command and bill at the same published rates.

Deploys that don't drop

Blue/green by default, automatic rollback on health-check failure, median deploy 51s.

GPU when you need it

L4 and A100 pools, per-second billing, scale to zero between jobs. Weights stream from your buckets free.

Agency-shaped

Projects, per-client roles, white-label status pages, one rolled-up invoice you can re-bill.