GPU Instance
SSH-ready Linux machine with NVIDIA GPU and CUDA. Install anything — Ollama, vLLM, training frameworks, or your own stack.
AI / ML
gpu
ssh
cuda
instance
vm
linux
gpu-instance
https://github.com/NVIDIA/nvidia-docker
Ready to deploy
Docker image
nvidia/cuda:12.8.1-devel-ubuntu24.04
Resources
8
CPU
32Gi
RAM
10Gi
Disk
Exposed ports
22 → 22 (public)
8080 → 80 (public)
Environment variables
SSH_PUBLIC_KEYrequired
Your SSH public key (ssh-rsa/ssh-ed25519). Injected into authorized_keys for passwordless login.
ROOT_PASSWORD
Root password for SSH (fallback if no public key). Leave empty to disable password auth.