Choose your path.

Bare Metal

Full control. Auto-detects GPU, installs drivers, CUDA, Docker, and the complete stack.

git clone https://github.com/ProjectZeroDays/FreeAI_AI-Inference-Workstation.git
cd FreeAI_AI-Inference-Workstation
sudo ./hardware/install-stack.sh
bash models/auto-download-models.sh

Docker Compose

Any host with NVIDIA Docker. Split or all-in-one profiles.

git clone https://github.com/ProjectZeroDays/FreeAI_AI-Inference-Workstation.git
cd FreeAI_AI-Inference-Workstation
docker compose --profile allinone up -d
# Dashboard: http://localhost:8030

Cloud & Live ISO

Kubernetes manifests, Vast.ai templates, RunPod kits, Hetzner bare-metal. Also bootable as a Live ISO.

docker pull ghcr.io/freeai/freeai:latest
# or boot freeaios-amd64.iso

Salad GPU

• Sign up at salad.com
• Select GPU instance type
• Use image: ghcr.io/freeai/freeai:latest
• Configure GPU profile in config/providers.json

Vast.ai

• Browse GPUs at vast.ai
• Select CUDA-capable instance
• Deploy using vastai/ template
• Auto-provisions FreeAI on first boot

RunPod

• Create pod at runpod.io
• Use FreeAI template or custom Docker image
• GPU-powered inference ready in minutes

Hostinger VPS

• Deploy Ubuntu 24.04 VPS
• SSH into server
• Run ./install.sh
• Configure reverse proxy in panel

DigitalOcean

• Create Ubuntu 24.04 droplet (8GB+ RAM)
• Run installation script
• Set up firewall rules
• Configure DNS for custom domain

Kubernetes

• Apply k8s/freeai-deployment.yaml
• Requires NVIDIA GPU operator
• Auto-scaling with VPA/HPA
• StatefulSet for model persistence