Intermediate

20–30 min

2 GB+ RAM

Deploy Perplexica on a RamNode VPS

Self-hosted AI search engine powered by SearXNG and large language models. Get cited, synthesized answers with no tracking, no per-query fees, and full control over your AI provider.

What Is Perplexica?

Perplexica is an open-source, self-hosted AI search engine inspired by Perplexity AI. It combines real-time web search through SearXNG with large language models to deliver cited, synthesized answers to natural language questions — with no tracking, no usage fees per query, and full control over which AI provider you use.

Note on naming: The project was recently rebranded from "Perplexica" to Vane on GitHub (at ItzCrazyKns/Vane). The community still widely uses "Perplexica," and the Docker images continue to work under both names.

Multiple search focus modes (Web, Academic, YouTube, Reddit)

Supports OpenAI, Anthropic, Groq, Ollama, and OpenRouter

Upload and query PDFs, TXT, and DOCX files

Self-hosted with full privacy — no external telemetry

Recommended RamNode Plan

Cloud LLM provider (OpenAI, Groq, Anthropic): A 2 GB RAM / 1 vCPU VPS is sufficient. All inference happens off-server.

Local LLM via Ollama: Plan for at least 8 GB RAM and ideally a GPU-enabled node. Smaller quantized models can squeeze onto a 4 GB instance with slow inference, but this is not recommended.

Prerequisites

A RamNode VPS with Ubuntu 22.04 LTS and root or sudo access
A domain name pointed at your VPS's IP address (an A record is sufficient)
Docker and Docker Compose installed (Docker installation guide)
Basic familiarity with the Linux command line
At least one LLM provider API key (OpenAI, Anthropic, or Groq) — or Ollama for local inference

Configure the Firewall

Open the ports needed for web traffic and SSH. Perplexica itself only needs to be reachable internally through nginx — do not expose ports 3000 or 8080 to the public.

Configure UFW

sudo ufw allow OpenSSH
sudo ufw allow 80/tcp
sudo ufw allow 443/tcp
sudo ufw enable

Single-Container Install (Recommended)

The simplest path is the bundled Vane image, which ships with SearXNG pre-configured inside the container.

Create working directory

mkdir -p /opt/perplexica
cd /opt/perplexica

Start the container

docker run -d \
  -p 127.0.0.1:3000:3000 \
  -v perplexica-data:/home/vane/data \
  --name perplexica \
  --restart unless-stopped \
  itzcrazykns1337/vane:latest

Binding to 127.0.0.1:3000 keeps the port off the public internet. The volume perplexica-data persists search history, uploaded files, and settings.

Verify

docker ps
docker logs perplexica

Wait about 30 seconds for the initial startup.

Docker Compose Install (More Control)

Use this method if you want to run SearXNG as a separate container for sharing across services or customizing independently.

/opt/perplexica/docker-compose.yml

version: "3.8"

services:
  searxng:
    image: searxng/searxng:latest
    container_name: perplexica-searxng
    volumes:
      - ./searxng:/etc/searxng
    networks:
      - perplexica-net
    restart: unless-stopped

  perplexica:
    image: itzcrazykns1337/perplexica:slim-latest
    container_name: perplexica
    environment:
      - SEARXNG_API_URL=http://searxng:8080
    ports:
      - "127.0.0.1:3000:3000"
    volumes:
      - perplexica-data:/home/perplexica/data
      - perplexica-uploads:/home/perplexica/uploads
    depends_on:
      - searxng
    networks:
      - perplexica-net
    restart: unless-stopped

volumes:
  perplexica-data:
  perplexica-uploads:

networks:
  perplexica-net:

Enable JSON format in SearXNG (required):

Create SearXNG config

mkdir -p searxng
cat > searxng/settings.yml << 'EOF'
use_default_settings: true

search:
  formats:
    - html
    - json
EOF

Start the stack

docker compose up -d
docker compose ps

Nginx Reverse Proxy & SSL

Install nginx and Certbot

sudo apt install nginx python3-certbot-nginx -y

/etc/nginx/sites-available/perplexica

server {
    listen 80;
    server_name search.yourdomain.com;

    location / {
        proxy_pass http://127.0.0.1:3000;
        proxy_http_version 1.1;
        proxy_set_header Upgrade $http_upgrade;
        proxy_set_header Connection "upgrade";
        proxy_set_header Host $host;
        proxy_set_header X-Real-IP $remote_addr;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_set_header X-Forwarded-Proto $scheme;
        proxy_read_timeout 120s;
        proxy_send_timeout 120s;
    }
}

The Upgrade and Connection headers are required — Perplexica uses WebSockets for streaming responses.

Enable site and obtain SSL

sudo ln -s /etc/nginx/sites-available/perplexica /etc/nginx/sites-enabled/
sudo nginx -t
sudo systemctl reload nginx

sudo certbot --nginx -d search.yourdomain.com

sudo systemctl status certbot.timer

Configure Your LLM Provider

Navigate to https://search.yourdomain.com in your browser. You will land on a setup screen where you configure your AI provider.

OpenAI — Enter your API key and select a model. gpt-4o-mini is cost-effective; gpt-4o gives noticeably better synthesis quality.

Groq — Generous free tier with very low latency. llama-3.3-70b-versatile is a strong choice.

Anthropic — claude-haiku-3-5 for speed, claude-sonnet-4-5 for quality.

Custom OpenAI-compatible — Works with OpenRouter (https://openrouter.ai/api/v1), LM Studio, or any compatible endpoint.

Ollama (local) — Set the API URL to http://host.docker.internal:11434. Pull your model first with ollama pull llama3.2:3b.

Settings are persisted to the Docker volume, so you only need to configure this once.

Keeping Perplexica Updated

Single-container method:

Update single container

docker pull itzcrazykns1337/vane:latest
docker stop perplexica && docker rm perplexica
docker run -d \
  -p 127.0.0.1:3000:3000 \
  -v perplexica-data:/home/vane/data \
  --name perplexica \
  --restart unless-stopped \
  itzcrazykns1337/vane:latest

Docker Compose method:

Update via Compose

cd /opt/perplexica
docker compose pull
docker compose up -d

Your data volume is not touched during an upgrade — search history and settings carry over.

Clean up old images

docker image prune -f

Search Modes & Features

Web — General web search — the default mode

Academic — Pulls from scholarly sources for research queries

YouTube — Searches and summarizes video content

Wolfram Alpha — Math, conversions, and factual lookups (requires API key)

Reddit — Surfaces community discussions

Speed / Balanced / Quality modes control how many sources are fetched and re-ranked before synthesis. Quality mode is slower but produces more thorough answers on complex topics.

File upload — Perplexica supports querying uploaded PDFs, TXT files, and DOCX documents directly. Uploaded files are stored in the persistent volume.

Troubleshooting

Blank page or cannot connect after setup

Check that nginx is running and the container is healthy:

Debug connectivity

sudo systemctl status nginx
docker ps
docker logs perplexica

If nginx shows a 502 error, the container is likely still starting. Wait 30–60 seconds and reload.

WebSocket errors / streaming responses cut off

Confirm your nginx config includes the Upgrade and Connection headers. Also check that proxy_read_timeout is set to at least 60 seconds.

SearXNG not returning results (Docker Compose setup)

The most common cause is JSON format not being enabled. Check your searxng/settings.yml and confirm json is listed under search.formats. Restart after changes:

Restart SearXNG

docker compose restart searxng

Disk usage growing over time

Search history and uploaded files accumulate in the Docker volume. Manage them through the Perplexica settings UI, or inspect directly:

Check disk usage

docker system df
docker volume inspect perplexica-data

Next Steps

Point your browser's built-in search to https://search.yourdomain.com/?q=%s for quick access from the address bar
Explore the Perplexica API at /api/search for integration with other tools
Consider adding HTTP basic auth in nginx if your instance is public but you want to restrict access

More Deployment Guides•View Cloud VPS Plans