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
Agent-run infrastructure · v4 control plane

Storage. Compute. DNS.
One plane, run by agents.

Managed S3 buckets, a Cloudways-grade compute engine, and Cloudflare-grade DNS — on one control plane where agents do the scaling, tiering, patching, and failover. Published rates. No egress toll between services.

S3 API v4 compatibleOCI imagesTerraform providerSOC 2 in audit
~/acme — n4n cli 4.2.1
$ n4n up --from ./acme
detected next.js · postgres · 1 bucket
provision fra1 · n4-std ×2 · 250 GB
dns acme.com · 41 records · dnssec
live https://acme.com in 1m 12s
6 agents attached
est. $132.40/mo
Live agent feedstreaming
right-sizecheckout 8→4 vCPU · −$61/mo
tier-cold318 objects → archive · −$1.21/mo
dns-failoverapi.acme.com → standby fra1 · 4s
cert-rotaterenewed 9 certs · 0 downtime
cost-guardegress ×3.4 · budget 61% used
99.99%
Uptime SLA, all tiers
14
Regions · 31 edge PoPs
38ms
p50 DNS resolve, global
51s
Median deploy, blue/green
$0
Egress between n4n services
01The stack

Three services that already know about each other.

No cross-account IAM. No per-product invoices. No egress toll between them. A bucket, an app, and a zone provisioned together share identity, region, and one budget.

01 / storage

Object storage

Byte-compatible S3 API — signature v4, multipart, presigned URLs, versioning, object lock. Keep your SDK and terraform; swap the endpoint.

standard$0.012 / GB-mo
egress$0.004 / GB
to n4n computefree
Explore storage
02 / compute

Compute engine

Git push to a managed runtime. OS patching, TLS, backups, blue/green cutovers, and autoscale policy are agent-owned. Root access if you want it.

n4-micro$6 / mo
n4-std$42 / mo
L4 gpu$0.62 / hr
Explore compute
03 / dns

DNS management

Anycast authoritative DNS across 31 PoPs. DNSSEC signed at creation, 4-second propagation, health-checked records that reroute themselves.

zonesunlimited, free
queries$0.30 / million
geodns$5 / zone
Explore DNS
02Watch it work

Write the command. The plane reacts.

input · cli
$ n4n storage create assets-prod --region fra1
↳ endpoint s3.fra1.n4n.host
$ n4n storage policy assets-prod public-read
$ aws s3 sync ./dist s3://assets-prod
1,243 objects · 412 MB · 9.1s
cdn warm in 31 PoPs
output · console
assets-prod
live
412 MB
Stored
1,243
Objects
$4.94
Projected month
100%
Cache hit, 24h
tier-cold agent
318 cold objects → archive
saved $1.21/mo · no config written
03Agent runtime

The ops work happens whether you're watching or not.

Six agents hold the pager across storage, compute, and DNS. Each one logs its reason, its cost delta, and a one-click revert. Nothing runs silently, nothing costs extra.

This month, across n4n
41,208
agent actions
$0
charged for them
−22%
median bill change
1.4%
actions reverted
Read how agents make decisions
right-size

Watches 14-day CPU and memory percentiles, proposes or applies a smaller class before renewal.

avg −18% compute spend
tier-cold

Moves untouched objects to Infrequent then Archive, writes the lifecycle rule, keeps the rollback.

avg −31% storage spend
dns-failover

Health-checks origins from 6 vantage points every 10s and swaps records to standby.

median swap 4s
cert-rotate

Renews TLS across apps, buckets, and custom domains. Rotates before the 30-day window.

0 expiry incidents
cost-guard

Tracks burn against budget, flags anomalies, drafts committed-use plans when usage is stable.

alerts at 60/80/95%
placement

Rebalances data off degraded hardware and steers GPU jobs to the nearest warm pool.

2.1 TB drained, 0 alerts
04Global network

14 regions. 31 PoPs. Live resolve times.

sampling every 2s
fra1GPU
Frankfurt
9ms
ams1
Amsterdam
12ms
lon1
London
10ms
iad1GPU
Ashburn
15ms
sfo1
San Jose
17ms
ord1
Chicago
13ms
gru1GPU
São Paulo
43ms
blr1
Bengaluru
19ms
sin1
Singapore
21ms
nrt1GPU
Tokyo
26ms
syd1
Sydney
36ms
jnb1
Johannesburg
47ms
dxb1GPU
Dubai
30ms
yyz1
Toronto
14ms
Replicas per object, same region, erasure coded
4s
DNS propagation across every PoP
100 Gbps
Per-region uplink, private backbone between regions
EU / US / IN
Residency pinning per bucket and per app
05Architecture

One control plane, three data planes.

Control plane · api.n4n.host
Auth, quotas, billing, audit log, terraform provider, CLI, webhooks
RESTgRPCS3 v4
storage plane
Erasure-coded object store, 3 replicas, edge cache in 31 PoPs, lifecycle engine.
s3.{region}.n4n.host
compute plane
Firecracker microVMs, OCI images, blue/green router, GPU pools, managed databases.
{app}.n4n.app
dns plane
Anycast authoritative servers, DNSSEC signer, health-check fleet, GeoDNS resolver.
ns1–ns4.n4n.host
Agent runtime
Reads every plane's telemetry, writes back through the same public API you use.
right-sizetier-colddns-failovercert-rotatecost-guardplacement
06Compare

What you'd otherwise stitch together.

CapabilityAWSCloudflareCloudwaysn4n.host
S3-compatible object storageyespartialnoyes
Internet egress per GB$0.09$0.00*$0.02$0.004
Egress between own servicesbilledfreebilledfree
Managed runtime + patchingadd-onnoyesyes
Autoscale to zeroyesn/anoyes
Anycast DNS + DNSSEC$0.50/zoneyesnofree
Health-checked failover$0.75/recpaid tierno$0.50/rec
Agent-driven right-sizingnononoincluded
One invoice for all threenononoyes
Published GPU hourly rateyesn/anoyes
* competitor terms as published Aug 2026 · figures illustrative
Before · four vendors

$247 / month

Buckets + egress$61
App servers + managed panel$132
DNS + WAF$25
Monitoring add-on$29
Four consoles, four support queues, one engineer keeping it glued.
After · n4n.host

$132.40 / month

Storage 900 GB, internal egress free$10.80
Compute 4× n4-std, autoscaled$118
DNS 9 zones, 12M queries$3.60
Agents, monitoring, backupsincluded
Measured on a customer migration in fra1. Run your own numbers.
07Migration

Move one workload. Keep the rest running.

01

Mirror, don't cut over

Point n4n storage mirror at your bucket and n4n dns import at your zone. Both stay authoritative on the old side.

day 0 · zero risk · ingress free
02

Run both, compare bills

Deploy the app beside its current host. Cost-guard reports the delta daily against your real traffic, not a calculator guess.

day 1–14 · first $500 credited
03

Flip the NS record

One record change moves traffic. A migration engineer sits on the call for anything above $500/mo, on your maintenance window.

day 15 · 4s propagation
08Reliability

The SLA is the same on the free tier.

Credits are automatic — we detect the breach from our own probes and apply it to your next invoice without a ticket.

See 90-day status
99.99%
Compute & storage SLA
100%
DNS resolution SLA
11×9
Object durability
auto
Breach credits, no ticket
09Pricing

Plans set the floor. Usage does the rest.

Full rate card + calculator
Hobby
$0/mo
10 GB storage · 1 n4-micro · unlimited zones · community support
Builder
$19/mo
100 GB storage · $10 compute credit · daily backups · 1-day support
Enterprise
Custom
Committed-use to −38% · private region · SSO/SCIM · 99.995% SLA
10FAQ

Questions operators actually ask.

Anything else, ask sales — a human answers within a business day.

Signature v4, multipart upload, presigned URLs, versioning, object lock, bucket policies, CORS, and lifecycle rules. Existing SDKs, aws s3, rclone, and Terraform's S3 provider work unmodified. Select and Glacier restore semantics are the two documented gaps.

Move one workload this week.

Free tier covers 10 GB storage, one instance, and unlimited DNS zones. Migration engineers are free above $500/mo — and the first $500 of compute is on us.

Start freeTalk to a migration engineer
no card · no sales call required · cancel by deleting the workspace