Private AI Server

Run LLMs on Your Own Server

Deploy private AI infrastructure in one click. Run Ollama + Open WebUI with Llama, Mistral and other open-source models on your own CPU cloud server. Your data never leaves your server.

Deploy AI Server Browse Images

From: $84.00/mo
Models: Up to ~32B
API: OpenAI-compatible
GPU: Not required

Plans

AI Server Plans

Virtualized CPU - Shared cores, great for most AI workloads. Best value for experimentation and moderate inference.

AI Starter

3B-7B models

$84/mo

CPU: 4 vCPU
RAM: 8 GB RAM
Storage: 100 GB NVMe
Models: 3B-7B models
Locations: 19 server locations
Access: Full root access

Deploy AI Starter

AI Business

AI Advanced

Max Power

Models up to ~32B

$263/mo

CPU: 12 vCPU
RAM: 32 GB RAM
Storage: 300 GB NVMe
Models: Models up to ~32B
Locations: 19 server locations
Access: Full root access

Deploy AI Advanced

One-Click Deploy

Pre-Built AI Images

Click to deploy. No configuration needed. Each app installs on Ubuntu 24.04 and is ready within minutes.

Ollama
Most Popular
Run open models locally. Llama, Mistral, Gemma, Phi - one command.
Deploy Ollama
Open WebUI
UI
ChatGPT-like interface for your local models. Multi-user, beautiful, and installed right alongside Ollama.
Deploy Open WebUI
OpenClaw
Agent
Your own AI assistant on Discord, Telegram and WhatsApp, with persistent memory and agent skills.
Deploy OpenClaw

Why self-host

Why Self-Host Your AI

100% Private: Your data never leaves your server. No API calls to third parties. Full privacy.
Any Model: Run Llama 3, Mistral, Phi, Gemma, or any GGUF/HF model that fits in RAM - up to ~32B parameters quantized. Switch models anytime.
One-Click Deploy: One-click install. Server + model runtime ready within minutes.
API Access: OpenAI-compatible endpoints. Use with any app, framework, or custom code.
Full Root Access: SSH in, install anything, customize everything. It's your machine.
Multiple Models: Run several models simultaneously on AI Business and AI Advanced plans.

DNS

DNS Configuration

Your server comes with a public IP. If you need a custom domain, configure these DNS records with your domain registrar.

A RecordRequired
Point your domain to the server IP
yourdomain.com → Server IP
AAAA RecordOptional
IPv6 address mapping (if available)
yourdomain.com → IPv6
CNAME RecordOptional
Alias for subdomains
api.yourdomain.com → yourdomain.com
Reverse DNS (rDNS)Optional
Maps IP back to your domain
Server IP → yourdomain.com

DNS changes are managed through your domain registrar. Propagation typically takes 5–30 minutes.

FAQ

Frequently Asked Questions

What models can I run?

Any open-source model that fits in your plan's RAM. AI Starter runs 3B-7B models (Phi, Llama 3 8B quantized). AI Business handles 8B-14B (Llama 3, Mistral). AI Advanced runs models up to ~32B parameters, quantized.

Do I need a GPU?

No. These are CPU cloud servers, and Ollama runs quantized models on CPU with no GPU required. For steady production workloads, switch to Dedicated CPU for guaranteed cores.

Is this OpenAI-compatible?

Yes. Ollama exposes an OpenAI-compatible API. Point any app that uses OpenAI's SDK at your server and it just works.

How is this different from AI VPS?

Same plans, same prices. This page is all about private LLMs with the Ollama + Open WebUI one-click app; AI VPS covers every AI use case, where you can also deploy OpenClaw, Hermes Agent, Mailcow or a clean OS image.

Your Models, Your Rules

No per-token fees. No rate limits. No data sharing. Run open models any time, for a flat monthly price.

Deploy AI Server - $84.00/mo