---
title: "Deploy Vane AI Search with SearXNG"
description: "Vane (formerly Perplexica) AI search with private SearXNG and login"
category: "AI/ML"
url: https://railway.com/deploy/vane-ai-search-with-searxng
---

# Deploy Vane AI Search with SearXNG

Vane (formerly Perplexica) AI search with private SearXNG and login

**[Deploy Vane AI Search with SearXNG on Railway](https://railway.com/template/vane-ai-search-with-searxng)**

Machine-readable deploy manifest (JSON, validated by TemplateCI): https://railway.com/deploy/vane-ai-search-with-searxng/manifest.json

- **Creator:** Anton Orel's Projects
- **Category:** AI/ML
- **Total deploys:** 1

## Template content

### Vane Gateway

- **Image:** caddy:2.11.9-alpine@sha256:003a97808e781aa9f7e75cc2f43d34d461099031faf5e20faefb9f61126836f0
- **Start command:** `/bin/sh -ec '[ "${#VANE_PASSWORD}" -ge 16 ] || { echo "VANE_PASSWORD must have 16 or more characters; refusing to start" >&2; exit 64; }; [ -n "${VANE_USERNAME:-}" ] && [ -n "${VANE_UPSTREAM:-}" ] && [ -n "${PORT:-}" ] || { echo "VANE_USERNAME, VANE_UPSTREAM and PORT are required" >&2; exit 64; }; VANE_PASSWORD_HASH="$(printf "%s\n" "$VANE_PASSWORD" | caddy hash-password)"; export VANE_PASSWORD_HASH; unset VANE_PASSWORD; printf "%s\n" "{" "admin off" "auto_https off" "persist_config off" "servers {" "trusted_proxies static private_ranges 100.64.0.0/10" "}" "}" ":{\$PORT} {" "handle /healthz {" "respond \"ok\" 200" "}" "handle {" "basic_auth bcrypt Vane {" "{\$VANE_USERNAME} {\$VANE_PASSWORD_HASH}" "}" "reverse_proxy {\$VANE_UPSTREAM} {" "header_up -Authorization" "flush_interval -1" "}" "}" "}" >/tmp/Caddyfile; exec caddy run --config /tmp/Caddyfile --adapter caddyfile'`
- **Health check:** /healthz
- **Public domain:** Yes

### Vane

- **Image:** itzcrazykns1337/vane:slim-v1.12.2@sha256:d2878cf9c91962aa3fc053b59bc9b89adcbdcaeb7ee36b54906e853464b2c190
- **Health check:** /api/providers

### SearXNG

- **Image:** searxng/searxng:2026.11.15-2d500cb6b@sha256:932f69dca010db9880bddebb2171f05d8e1a25d04a59149ff65409993d680024
- **Start command:** `/bin/sh -ec 'printf "%s\n" "use_default_settings: true" "general:" "  instance_name: Vane SearXNG" "search:" "  formats:" "    - html" "    - json" "server:" "  limiter: false" "  public_instance: false" "  image_proxy: false" "engines:" "  - name: brave" "    disabled: true" >/etc/searxng/settings.yml; exec /usr/local/searxng/entrypoint.sh'`
- **Health check:** /healthz

## Documentation

# Deploy and Host Vane AI Search with SearXNG on Railway

Deploy Vane, the open-source AI answering engine formerly called Perplexica, with its own private SearXNG metasearch instance and a password-protected gateway. Ask a question and get an answer with numbered citations linking to the web sources it used.

## About Hosting Vane AI Search with SearXNG

[Vane](https://github.com/ItzCrazyKns/Vane) searches the web through [SearXNG](https://github.com/searxng/searxng), reads the results, and has the language model of your choice write a cited answer. This template runs Vane `v1.12.2` and SearXNG as separate private services, each pinned to an exact image digest, and puts a Caddy gateway with a generated password in front of them. Vane's settings, chat history and uploads are kept on a persistent volume.

## Why Deploy Vane AI Search with SearXNG on Railway

Vane has no login of its own and stores your model API keys, so exposing it directly would let anyone read them and spend your credits. This template keeps Vane and SearXNG on Railway's private network, makes the authenticated gateway the only public entry point, and generates every credential at deploy time. It uses the current Vane release rather than the retired Perplexica image.

## Common Use Cases

- A private Perplexity-style research assistant using your own OpenAI, Anthropic, Gemini, Groq, OpenRouter or Ollama key.
- Cited web answers for a small team without sending searches to a third-party search API.
- Chat with uploaded documents alongside web results, with history kept between sessions.

## Dependencies for Vane AI Search with SearXNG Hosting

You need an API key or endpoint for a language model provider. The template includes everything else and generates all credentials.

### Deployment Dependencies

- Vane `v1.12.2` (slim image) for the web app and answer engine.
- SearXNG for private metasearch, configured for JSON output and private use.
- Caddy `2.11.9` for HTTP Basic Auth and the public reverse proxy.
- One 5 GB volume for Vane's settings, SQLite chat history and uploads.

After deployment, open `Vane Gateway` and sign in with `VANE_USERNAME` and `VANE_PASSWORD` from that service's variables. In Vane's setup screen, add your model provider and key, then pick a chat model and an embedding model.

This is a single-replica deployment with one shared login, intended for personal or small-team use. Public search engines can rate-limit traffic from cloud IP addresses at times. To enable SearXNG's Wikipedia, Wikidata and OpenStreetMap engines, set `SEARXNG_CONTACT_URL` on the `SearXNG` service to your own contact URL or email address.


## Similar templates

- [Chat Chat](https://railway.com/deploy/-WWW5r) — Chat Chat, your own unified chat and search to AI platform.
- [Weaviate](https://railway.com/deploy/weaviate-1) — Vector database with hybrid search and an API key-protected API
- [Möbius Personal AI OS](https://railway.com/deploy/mobius-v01-personal-ai-os) — Self-hosted AI agent that builds and runs apps for you, with memory.

Open this page in a browser: https://railway.com/deploy/vane-ai-search-with-searxng
