One private chat window for ChatGPT, Claude, Gemini and models on your own computer.
Vane (formerly Perplexica)
Ask a question, get a written answer with sources, from a search engine you run yourself.
AI helpers
MIT
What it does
Vane (until recently called Perplexica) works like Perplexity: you ask a question in plain words, it searches the web privately through its own built-in SearXNG, reads the pages and writes an answer with numbered sources. It can also search discussions or academic papers, look at files you upload, and show images, videos, weather and news. Use a free local model through Ollama or your own key for OpenAI, Claude, Gemini or Groq.
- Written answers with numbered sources, from a private search engine
- Works with free local models or your own AI keys
- Quick, balanced or deep research modes; web, discussions or academic papers
- Upload files and ask about them; images, videos, weather and news cards
Great for: People who like Perplexity-style answers but want their searches private and on their own computer.
- LicenceMIT
- Where it runsOnline (about $5/month, usage-based) · Your computer (free)
- Built withtypescript
- Your dataStays in your own copy
- SourceItzCrazyKns/Vane
Ways to use it
Choose one when you tell your AI. It sets everything up with you, in plain words.
- Online with Railway about $5/month, usage-basedA general-purpose host that can run any of these apps, including your own customised version. Usage-based pricing with a free trial. Third-party host.
- On my own computerFreeYour AI installs what it needs and explains each step. It works while your computer is on.
Vane (formerly Perplexica) has no login by default: keep it private, or ask your AI to add a password before putting it online.
Hosts are independent companies, not BespokeAI. You sign up and pay them directly. We may earn a referral fee, which never changes your price.
How people are using this project
Examples of what people ask their AI to do with Vane (formerly Perplexica). Start from one, or just tell your AI and it will ask what you'd like.
-
“Set Vane up on my laptop with a free local model and make it the default search engine in Firefox.”
Start with this idea -
“I'm in the UK. Make answers short, in British English, metric, and prefer NHS, BBC and gov.uk sources where they fit.”
Start with this idea -
“My 15-year-old uses this for school. Make it explain clearly, always show sources, and prefer academic and encyclopaedia sources.”
Start with this idea -
“Make an iPhone Shortcut that sends a spoken question to my Vane at home and reads the answer back to me.”
Start with this idea -
“When I search for a recipe, show a card at the top with ingredients and cooking time, like the weather card.”
Start with this idea -
“I want to reach Vane from outside the house. Add a simple password screen so only my family can use it.”
Start with this idea
Can I trust it?
- MIT. Very permissive. You can use, change and share it, even commercially, as long as you keep the copyright notice. No obligation to publish your changes.
- Your data: stores personal data, on your own device or server. Search history and chats in SQLite (data/db.sqlite), settings and API keys in data/config.json, and uploaded files under the data folder. In Docker all of this is in the vane-data volume (/home/vane/data).
- Other services: SearXNG (bundled in the Docker image) (required, free, self-hosted); Chat and embedding models (OpenAI, Anthropic, Gemini, Groq, or OpenAI-compatible) (optional, pay-per-use with your own key); Ollama, LM Studio or Lemonade (optional, free (uses your own computer)); Local Transformers embeddings (optional, free)
- Pinned version
348feca. Every handover uses exactly this code, which we've reviewed. - Vane has no login at all. Anyone who can reach port 3000 can use your AI keys and read your search history. Keep it on your home network, or put it behind a reverse proxy with a password and HTTPS.
- Your question and the pages found are sent to the AI provider you choose. Use Ollama or another local model to keep everything at home; the web searches themselves go out through SearXNG to engines like Google, Bing and DuckDuckGo without your identity.
- API keys are stored in plain text in data/config.json inside the vane-data volume. Don't share that volume or paste it into an AI chat.
- Development is slower in 2026 (one main maintainer), so check for security updates before exposing it.
Did it work for people?
No reports yet
Be the first to tell others how it went.
More for ai helpers
Ask questions about your own documents with an AI that can run entirely on your computer.
A personal AI that knows your notes and documents and can research the web for you.