Skip to main content
Inferix

Host Setup Quick Start

Fast track to becoming a GPU provider. Get from signup to live in 2-4 hours.

5-Minute Overview

  1. 1.Create a hosting account → Settings → Account Type → Accept hosting agreement
  2. 2.Check system requirements → Ubuntu 18.04+, Nvidia GPU 10-series+, 10+ Mbps internet
  3. 3.Install Ubuntu & drivers → Download Ubuntu 22.04 → Install Nvidia driver
  4. 4.Run bootstrap script → One command auto-detects hardware and registers machine
  5. 5.Configure ports → Open 100+ ports on router, verify with port checker
  6. 6.List machine → Set price and availability → Accept client rentals

Total time: 2-4 hours

System Checklist

GPU

Nvidia 10-series or newer (RTX 3060+, T4, A10, H100)

CPU

At least 1 core per GPU, supports AVX instruction set

RAM

4 GB minimum per GPU (8+ GB recommended)

Storage

SSD/NVMe, ≥128 GB per GPU, XFS partition

Network

≥10 Mbps per machine, stable connection, ≥99% uptime

OS

Ubuntu 18.04+ (22.04 LTS recommended)

Step 1: Prepare Your Account

Go to Settings → Account Type

  1. Review and accept the Hosting Agreement
  2. Click Convert to Hosting Account
  3. Your account will be converted immediately

Next: Join our Discord and link your hosting account to #host-support channel

Step 2-3: Install OS & Drivers

Download Ubuntu 22.04 LTS

https://ubuntu.com/download/server

Install Nvidia Driver

sudo apt update && sudo apt upgrade -y
sudo apt install nvidia-driver-550 -y
nvidia-smi -q

Install Docker & Nvidia Runtime

sudo apt install docker.io nvidia-docker2 -y
docker run --rm --gpus all nvidia/cuda:11.0-runtime nvidia-smi

Step 4: Run Bootstrap Script

🔑 Get Your API Token

Call this endpoint with your JWT token:

curl -H "Authorization: Bearer <YOUR_JWT>" \
  https://api.platform.dev/api/v0/providers/me/api-token

📥 Run Bootstrap Script

export HF_API_URL=https://api.platform.dev
export HF_TOKEN=<your_access_token>
export HF_REGION=US-East
export HF_COUNTRY=US
export HF_PRICE_PER_GPU_HOUR=0.65

curl -fsSL https://platform.dev/scripts/vast-provider-bootstrap.sh | bash

Step 5: Configure Network & Ports

Router Port Forwarding

  1. Log into your router admin panel
  2. Find Port Forwarding section
  3. Forward 16000-16399 → your machine's local IP
  4. Apply changes

Test & Configure Daemon

nc -l -p 16000

# Visit portchecker.co with your public IP and port 16000
# Once verified, inform daemon:
echo -n "16000-16399" | sudo tee /var/lib/vastai_kaalia/host_port_range

Step 6: List Your Machine

  1. Go to Machines page
  2. Find your registered machine (wait 1-5 minutes)
  3. Click List button
  4. Set hourly price (e.g., $0.65 per GPU)
  5. Set availability end date
  6. Click Confirm Listing

✅ Your machine is now accepting rentals!

Important Reminders

  • ✗ Don't host from a client account (must convert first)
  • ✗ Don't shut down or reboot while clients rent your machine
  • ✗ Don't run competing workloads during rentals
  • ✓ Test before listing: `vastai self-test machine <ID>`
  • ✓ Disable auto-updates: `sudo apt-mark hold nvidia-driver-*`

Machine Verification Timeline

Your machine starts as Unverified. It becomes verified automatically based on:

  • ≥99% uptime
  • Passing performance benchmarks
  • Reliable job completion
  • Stable network connectivity

Timeline: 24-48 hours (if all criteria met)

Ready to become a provider?