# Perplexica + SearXNG

Cited answers you host. Perplexica is MIT; SearXNG is AGPL-3.0 — Perplexity without the answer in someone else's log.

**Replaces:** Perplexity

**Tier:** Medium

## Problem

Perplexity is a hosted answer engine: search plus a model plus citations. You wanted that loop on hardware you run. Perplexica is MIT (cited answers). SearXNG is AGPL-3.0 (the search backend). You still bring a model (Ollama or a cloud key). This is not Perplexity's every focus mode or Pro index.

## Stack

- [Perplexica](https://github.com/ItzCrazyKns/Perplexica) — MIT self-hosted cited-answer engine
- [SearXNG](https://docs.searxng.org) — AGPL-3.0 metasearch backend Perplexica queries

## Estimated SaaS cost saved

| You stop paying | Typical / year | Self-host instead |
|---|---:|---|
| Perplexity paid (small team) | $1,200 | Perplexica + SearXNG + your model |

**About $1,200 / year** on a small team, before you count lock-in.

## Deploy on Railway

Placeholder: Perplexica + SearXNG (JSON enabled) plus your model endpoint. No one-click template in this repo yet — Start on Railway. Plain chat stays on /stacks/open-webui-litellm.

[Start on Railway](https://c17.io/go/railway/perplexica-search?from=%2Fstacks%2Fperplexica-search.md)

[Open WebUI if this is chat, not cited search](https://c17.io/stacks/open-webui-litellm)

## AGENT_PROMPT

```
Replace Perplexity with Perplexica plus SearXNG.

- Enable JSON on SearXNG. Point Perplexica at it and at a model (Ollama or LiteLLM). Persist config.
- Ask one question you actually ask Perplexity. Confirm citations. Write AGENT_PROMPT.md.
- Do not claim Perplexity Pro indexes or every focus mode.
```

## FAQ

### Why not just Open WebUI?

Open WebUI is chat (/replace/chatgpt). Perplexity is search-then-answer with citations. Perplexica is that job.

### Do I still pay a model?

If you route to a hosted model, yes. Local Ollama is the other path. SearXNG is the search, not the LLM.

Source: https://c17.io/stacks/perplexica-search