Vane
ActiveDescription
An AI-powered answering engine with multi-model integration, web search and local knowledge base, providing a Perplexity-like search experience.
Key Features
- Multi-model LLM support — Use local models via Ollama or connect to OpenAI, Anthropic Claude, Google Gemini, Groq and more
- SearxNG privacy search — Integrates SearxNG meta search engine aggregating multiple search providers without tracking identity
- Smart search modes — Speed Mode for quick answers, Balanced Mode for daily searches, Quality Mode for deep research
- Multi-source retrieval — Web search, academic papers, discussions; can restrict to specific domains
- File upload Q&A — Upload PDFs, text, images and ask questions about document content
- Widgets info cards — Weather, calculations, stock prices and other quick-lookup widgets auto-displayed
Use Cases
💡 Privacy-focused daily search — Users who don't want search history tracked, using local LLMs for fully offline Q&A
💡 Academic literature research — Search academic papers and use LLM to summarize and answer questions about paper content
💡 Technical documentation lookup — Restrict search to specific documentation sites for precise answers
💡 Local file analysis — Upload PDFs or documents and perform in-depth Q&A on file contents
Strengths & Limitations
✅ Strengths
- • Actively maintained, recent updates
- • High community interest (36.6k stars)
- • Permissive open-source license (MIT)
- • Established track record (2 years in production)
Tags
Categories
Quick Start
docker run -d -p 3000:3000 -v vane-data:/home/vane/data --name vane itzcrazykns1337/vane:latest