Skip to main content
This feature is in public preview.
We’ve switched to a new API. Migrate your nodes →
This guide walks you through launching your first GPU instance on SF Compute.

​
Install the CLI

curl -fsSL https://cli.sfcompute.com | bash

​
Log in

sf login
This opens your browser to authenticate and stores your credentials locally. You can check your credit balance anytime with sf billing balance.

​
Check what you can buy

Hardware is described by instance SKUs. Preview what’s currently for sale, one row per SKU. 
sf availability

INSTANCE SKU             NEXT AVAILABLE  NEXT 48H
Amsterdam H100 (IB)      now (24)        ▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▁▁▁▁▁  peak: 24
Ashburn H100 (IB)        now (64)        ▇▇▇▇▇▇▇▇▇▇▁▁▁▁▁▁▁▁▁▁  peak: 64
Chicago A100 (Ethernet)  now (96)        ▇▇▇▇▇▄▄▄▄▄▂▂▂▂▂▁▁▁▁▁  peak: 96
Seattle H100 (IB)        now (128)       ▇▇▇▇▇▇▇▇▇▇▁▁▁▁▁▁▁▁▁▁  peak: 128
Tokyo H100 (IB)          now (48)        ▇▇▇▇▇▇▇▇▇▇▇▇▁▁▁▁▁▁▁▁  peak: 48
NEXT AVAILABLE is the soonest a node frees up and how many — now (N), a time like Today, 8:00pm (4), or — when nothing’s scheduled; NEXT 48H sparklines the schedule with the peak shown to the right.

​
Create an empty pool

A pool tracks your compute allocation over time. You buy compute time into a pool with orders, and instances on the pool run when there’s allocated time. 
sf pools create --name dev

​
Buy compute time

This will charge your account. Make sure you have credits — check with sf billing balance or top up on the dashboard.
Place a buy order for compute time on your pool. The command guides you through each option interactively, including number of instances, start time, duration, rate, and the instance SKU the order pins to. 
sf orders create --pool dev --side buy

​
Check your order

sf orders list

SIDE  STATUS  PERIOD   NODES  RATE             POOL      ID
buy   filled  Mar 23       1  ≤$20.00/node/hr  dev       ordr_...
If the order shows cancelled, the price was too low or there was no availability. Try a higher rate, a later start time, or use --allow-standing to leave the order on the book.

​
List images

Images are bootable OS disk images. You’ll need an image name or ID to create an instance. 
sf images list

NAME                      VISIBILITY  STATUS     CREATED
ubuntu-22.04.5-cuda-12.7  public      completed  Feb 23, 6:50pm
SFC provides public images with Ubuntu and CUDA pre-installed. You can also build and upload your own custom images.

​
Create an instance

First, create a startup script that injects your SSH key so you can connect to the instance. 
cat >startup.sh <<SCRIPT
#!/bin/bash

mkdir -p /root/.ssh
cat >>/root/.ssh/authorized_keys <<"EOF"
$(cat ~/.ssh/id_rsa.pub 2>/dev/null)
$(cat ~/.ssh/id_ecdsa.pub 2>/dev/null)
$(cat ~/.ssh/id_ecdsa_sk.pub 2>/dev/null)
$(cat ~/.ssh/id_ed25519.pub 2>/dev/null)
$(cat ~/.ssh/id_ed25519_sk.pub 2>/dev/null)
$(cat ~/.ssh/id_xmss.pub 2>/dev/null)
$(cat ~/.ssh/id_dsa.pub 2>/dev/null)
EOF
SCRIPT
Then create an instance on your pool with the public image. 
sf instances create --pool dev --image ubuntu-22.04.5-cuda-12.7 --cloud-init ./startup.sh
Your pool already has compute time from the order you placed. Since the order’s start time was now, the instance will be allocated within about a minute and begin booting. Compute ownership and instances are separate: you can terminate and recreate instances without losing your purchased compute time.

​
Watch your instance start

sf instances list
The image download and boot process takes up to 10 minutes before SSH is available.

​
Connect via SSH

sf instances ssh {instance-name}
Verify the GPUs are available: 
nvidia-smi

​
Next steps

You now have a running GPU instance. Here are some things to try: Sell back unused compute. Place a sell order as a standing order — if someone buys it, you get credits back and the instance loses its compute time. 
sf orders create --pool dev --side sell --nodes 1 \
  --start now --duration 1h --min-rate 8.00 --allow-standing
Check your pool’s allocation schedule to see the compute time you own. 
sf pools get dev
Learn more:
  • Pools — manage compute allocation
  • Orders — buy and sell compute time
  • Instances — SSH, logs, images, and lifecycle