Vane

Why we included this project

Vane turns web results into conversational answers with citations, and it runs on your own hardware, so you get the convenience of a Perplexity-style assistant without handing your queries to a third party. It works with local models through Ollama, and can also use OpenAI, Claude, Gemini, or Groq, so you can pick a different provider per search. Speed, Balanced, and Quality modes trade latency for depth, and you can restrict results to the web, discussions, or academic papers depending on the task. File uploads let you ask questions about your own PDFs and documents, and the bundled SearxNG integration keeps your identity private across multiple engines. If you want a private, self-hosted alternative to commercial AI search, this is a complete package.