sumi cloud
Login

Server noise without the ops overhead

sumi polls your servers every minute and streams CPU, memory, disk, and network straight to your terminal. Add cloud history (30 or 90 days) and Telegram alerts only if you need them — no extra services, no dashboards to babysit.

Open source CLI · Cloud plans from €49/year for history + Telegram alerts

Install the free CLISee pricing ↓
sumi
╭─ THERMAL ──────────────────────────╮ ╭─ CPU ───────────────────────────────╮
│ temp  55.2°C                       │ │ load █████████████░░░░░░░░░░░░░░░   │
│ arm   1800 MHz                     │ │ cores 4                             │
╰────────────────────────────────────╯ ╰─────────────────────────────────────╯

╭─ Network ──────────────────────────╮ ╭─ Top Processes ────────────────────╮
│ Iface: eth0                        │ │ NAME      CPU%  MEM%               │
│ Rx: 124.6 KB/s                     │ │ chromium  38.0   5.2               │
│ Tx:  18.3 KB/s                     │ │ code       8.2   3.4               │
╰────────────────────────────────────╯ ╰────────────────────────────────────╯

One dashboard for the whole fleet, with history and alerts

This is what the paid plans unlock: one place to check every device, 30 or 90 days of history for debugging, and Telegram alerts before a user notices first.

Signal grid

CPU

42%

Warn 70% · Danger 88%

Memory

68%

Warn 75% · Danger 90%

Disk used

57%

Inodes 21%

FD used

19%

Open conn 143

Disk await

1.8 ms

Warn 8 ms · Danger 20 ms

Packet loss

0%

Latency 8.2 ms

Temperature

54C

Warn 75C · Danger 85C

Retransmits

2

Network looks healthy

CPU % + Memory %

FD used % + Open connections

Pricing

Free
EUR0

For a single machine you want to watch locally without paying for cloud yet.

  • One device in the terminal UI with no subscription pressure.
  • Works on macOS, Linux, and Raspberry Pi with the same binary.
  • Useful when you want better visibility than htop, not a hosted dashboard.
Get the free CLI
Popular
Hobby
EUR49/year

For side projects that now deserve history and alerts.

  • Monitor up to 5 devices from one shared cloud dashboard.
  • Keep 30 days of history so you can inspect a problem that started last week.
  • Get Telegram alerts before a user or friend notices first.
Start Hobby
Pro
EUR99/year

For a small production fleet you still operate yourself.

  • Monitor up to 25 devices without switching tools again too early.
  • Keep 90 days of history for regressions, slow drift, and repeat incidents.
  • Stay on a simple workflow even when the fleet stops being simple.
Choose Pro

How it works

Install with one command

A pre-built static binary for macOS, Linux, and Raspberry Pi. No runtime dependencies, no package manager required.

curl -sL https://getsumi.dev/install.sh | sh -s -- YOUR_TOKEN

Add cloud when the fleet grows up

Use the cloud layer for 30 to 90 day history, a shared dashboard, and Telegram alerts when live-only monitoring stops being enough.

FreeHobbyPro
PriceFreeEUR 49/yearEUR 99/year
Devices1525
History—30 days90 days
Cloud dashboard—✓✓
Telegram alerts—✓✓
System metrics✓✓✓
Network metrics✓✓✓
Storage metrics✓✓✓
Process list✓✓✓
System events—✓✓
WireGuard monitoring✓✓✓
NDJSON streaming✓✓✓
Create free account

Who sumi is for

Your monitoring is a bunch of open SSH tabs.

sumi replaces that habit with one view across all devices, refreshed every minute without touching your servers.

You've used Grafana. You know what it costs to maintain.

sumi is one binary per host. No Prometheus, no storage decisions, no dashboards to babysit — just metrics where you already look.

You want to know before your users do.

Set a threshold on any metric, link Telegram once. sumi sends a message when something needs attention, not when CPU spikes for two seconds.

You'd rather fix incidents than build monitoring pipelines.

EUR 49/year for 30-day history and Telegram alerts on up to 5 servers. The alternative is a self-hosted stack you'd also have to maintain.