Tabby Self-Hosted AI Coding Assistant on RamNode VPS

What Tabby Offers

Code Completion: Real-time, context-aware suggestions supporting 80+ languages
Answer Engine: Integrated Q&A that understands your codebase and docs
Inline Chat: AI assistant inside your editor for refactoring, explanations, and reviews
Repository Context: Connect Git repos for more relevant suggestions
Multi-IDE Support: VS Code, JetBrains, Vim/Neovim, and Emacs
Enterprise Features: Team management, SSO, usage analytics, and audit logging

Prerequisites & VPS Sizing

Requirements

RamNode Account — up to $500 in annual credits for new accounts
SSH client (Terminal, PuTTY, or Windows Terminal)
Domain name (optional) — e.g. tabby.yourdomain.com for HTTPS
IDE: VS Code, JetBrains, Vim/Neovim, or Emacs

Model Size → VPS Plan

Model Size	Min RAM	Min CPU	Recommended Plan
0.5B – 1.5B	4 GB	2 vCPUs	Premium KVM 4GB ($24/mo)
3B	8 GB	4 vCPUs	Premium KVM 8GB ($48/mo)
7B	16 GB	4+ vCPUs	Premium KVM 16GB ($96/mo)
13B – 14B	32 GB	8 vCPUs	Premium KVM 32GB ($192/mo)

💡 Cost Savings: New RamNode accounts receive up to $500 in annual credits. A Premium KVM 8GB plan running a 3B model provides excellent quality for small-to-medium teams and is fully covered by credits for over a year.

Provision & Prepare Your VPS

Provision an Ubuntu 22.04 or 24.04 LTS VPS from RamNode, then SSH in:

SSH into your server

ssh root@YOUR_SERVER_IP

Update system packages

apt update && apt upgrade -y
apt install -y curl wget git ufw

Install Docker

Install Docker Engine

# Add Docker's official GPG key
curl -fsSL https://download.docker.com/linux/ubuntu/gpg \
  | gpg --dearmor -o /usr/share/keyrings/docker-archive-keyring.gpg

# Add the Docker repository
echo "deb [arch=$(dpkg --print-architecture) \
  signed-by=/usr/share/keyrings/docker-archive-keyring.gpg] \
  https://download.docker.com/linux/ubuntu \
  $(lsb_release -cs) stable" | tee /etc/apt/sources.list.d/docker.list

# Install Docker Engine
apt update
apt install -y docker-ce docker-ce-cli containerd.io docker-compose-plugin

# Verify installation
docker --version

Configure Firewall

Set up UFW

ufw allow OpenSSH
ufw allow 8080/tcp
ufw enable

💡 HTTPS: If exposing Tabby publicly, place it behind a reverse proxy with TLS. See the HTTPS with Caddy section below.

Deploy with Docker Compose

Create data directory

mkdir -p /opt/tabby/data

Create /opt/tabby/docker-compose.yml

version: '3.5'

services:
  tabby:
    image: registry.tabbyml.com/tabbyml/tabby
    container_name: tabby
    restart: always
    command: >
      serve
      --model StarCoder-1B
      --chat-model Qwen2-1.5B-Instruct
      --device cpu
    volumes:
      - "/opt/tabby/data:/data"
    ports:
      - "8080:8080"
    environment:
      - TABBY_WEBSERVER_JWT_TOKEN_SECRET=CHANGE_ME_TO_RANDOM_STRING

⚠️ Security: Replace CHANGE_ME_TO_RANDOM_STRING with a strong secret. Generate one with: openssl rand -hex 32

Start Tabby

cd /opt/tabby
docker compose up -d

# Monitor first launch (model download may take several minutes)
docker compose logs -f tabby

Wait until you see a message indicating the server is listening on port 8080.

Initial Admin Setup

Open your browser and navigate to http://YOUR_SERVER_IP:8080. On first access, Tabby will prompt you to create an admin account. Complete the registration to access the admin dashboard where you can manage users, configure repository context, and monitor usage statistics.

Model Selection Guide

Completion Models

Real-time code suggestions as you type:

Model	Size	Best For	License
Qwen2.5-Coder-1.5B	1.5B	Low-resource deployments	Apache 2.0
StarCoder2-3B	3B	Balanced quality/speed	BigCode-OpenRAIL-M
Qwen2.5-Coder-7B	7B	High-quality completions	Apache 2.0
DeepSeekCoder-6.7B	6.7B	Python/JS focused	Deepseek License
CodeLlama-13B	13B	Maximum quality (high RAM)	Llama 2

Chat Models

Power the Answer Engine and inline chat:

Model	Size	Notes	License
Qwen2-1.5B-Instruct	1.5B	Lightweight default	Apache 2.0
Qwen2.5-Coder-7B-Instruct	7B	Code-specialized chat	Apache 2.0
Qwen3-8B	8B	Latest generation quality	Apache 2.0
CodeGemma-7B-Instruct	7B	Google's code model	Gemma License

Swap models

# Update --model and --chat-model in docker-compose.yml, then:
cd /opt/tabby
docker compose down && docker compose up -d
# Tabby will automatically download the new model on startup

IDE Extension Setup

VS Code

Open Extensions marketplace → search "Tabby" → install by TabbyML
Settings (Ctrl+,) → search "Tabby" → set Server Endpoint to http://YOUR_SERVER_IP:8080
Status bar shows a Tabby icon when connected — start typing to see suggestions

JetBrains IDEs

Settings → Plugins → Marketplace → search "Tabby" → install and restart
Settings → Tools → Tabby → enter server endpoint URL

Vim / Neovim

vim-plug setup

" Add to your plugin manager
Plug 'TabbyML/vim-tabby'

" Configure endpoint in vimrc or init.lua
let g:tabby_server_url = 'http://YOUR_SERVER_IP:8080'

Configure Repository Context

Tabby can index your Git repositories for more contextually relevant completions — one of its most powerful features for teams.

Via Admin UI

Log in to the admin dashboard at http://YOUR_SERVER_IP:8080
Navigate to Repositories → add Git URLs (GitHub, GitLab, or self-hosted)
Tabby clones and indexes repositories in the background

Via Configuration File

/opt/tabby/data/config.toml

[[repositories]]
name = "my-project"
git_url = "https://github.com/your-org/your-repo.git"

[[repositories]]
name = "internal-api"
git_url = "https://git.yourdomain.com/team/internal-api.git"

Trigger indexing

docker exec tabby /opt/tabby/bin/tabby-cpu scheduler --now

Optional: HTTPS with Caddy

Install Caddy

apt install -y debian-keyring debian-archive-keyring apt-transport-https
curl -1sLf https://dl.cloudsmith.io/public/caddy/stable/gpg.key \
  | gpg --dearmor -o /usr/share/keyrings/caddy-stable-archive-keyring.gpg
curl -1sLf https://dl.cloudsmith.io/public/caddy/stable/debian.deb.txt \
  | tee /etc/apt/sources.list.d/caddy-stable.list
apt update && apt install -y caddy

Create /etc/caddy/Caddyfile

tabby.yourdomain.com {
    reverse_proxy localhost:8080
}

Enable and open firewall

systemctl reload caddy

ufw allow 443/tcp
ufw allow 80/tcp

Caddy automatically obtains and renews Let's Encrypt certificates. Update your IDE extensions to use https://tabby.yourdomain.com.

Maintenance & Performance Tuning

Update Tabby

Pull latest version

cd /opt/tabby
docker compose pull
docker compose down
docker compose up -d

Backup Strategy

Daily compressed backup

tar czf /backups/tabby-$(date +%F).tar.gz /opt/tabby/data

Monitoring

Check health and resources

# Container status
docker compose ps

# Live logs
docker compose logs -f tabby

# Resource usage
docker stats tabby

Performance Tuning

Parallelism: Add --parallelism 4 to the serve command to handle concurrent requests (match to your vCPU count).

Add swap space (recommended for larger models)

fallocate -l 4G /swapfile
chmod 600 /swapfile
mkswap /swapfile
swapon /swapfile
echo "/swapfile none swap sw 0 0" >> /etc/fstab

Troubleshooting

Container Fails to Start (OOM)

Your VPS may lack RAM for the selected model. Try a smaller model (e.g., StarCoder-1B) or upgrade your plan.

Slow Completions (>2–3 seconds)

Model too large for CPU. Switch to a smaller model, stop competing processes, and check swap usage with free -h.

IDE Cannot Connect

Verify container is running (docker compose ps), port 8080 is open (ufw status), and you can reach the web UI in a browser.

Model Download Fails

Pre-download the model before starting:

Pre-download model

docker run --rm -v /opt/tabby/data:/data \
  --entrypoint /opt/tabby/bin/tabby-cpu \
  registry.tabbyml.com/tabbyml/tabby \
  download --model StarCoder-1B

Quick Reference

Essential Commands

Action	Command
Start Tabby	`docker compose up -d`
Stop Tabby	`docker compose down`
View logs	`docker compose logs -f tabby`
Update	`docker compose pull && docker compose up -d`
Check status	`docker compose ps`
Monitor resources	`docker stats tabby`
Index repos	`docker exec tabby /opt/tabby/bin/tabby-cpu scheduler --now`

Key URLs

Web UI: http://YOUR_SERVER_IP:8080
API Docs: http://YOUR_SERVER_IP:8080/swagger-ui
Health Check: http://YOUR_SERVER_IP:8080/v1/health
Tabby Docs: tabby.tabbyml.com/docs
Model Registry: tabby.tabbyml.com/docs/models

Tabby Deployed Successfully!

Your self-hosted AI coding assistant is now running. Connect your IDE and enjoy private, context-aware code completions powered by your own infrastructure.

More Deployment Guides →View Cloud VPS Plans →

Deploy Tabby AI Coding Assistant