Perplexica (now renamed to Vane) is a self-hosted answering engine that combines LLM responses with web search. It is designed as an open-source alternative to hosted AI search tools. The Docker image bundles SearxNG for search, making it easy to run a full stack with a single command. Perplexica is typically deployed for private search and research workflows.
Note: The project has been renamed to Vane. The repository is now at https://github.com/ItzCrazyKns/Vane
- Search + LLM answering engine
- Bundled SearxNG integration
- Docker-based quickstart
- Configurable API keys for models
- Support for local LLMs (Ollama) and cloud providers
- Smart search modes (Speed, Balanced, Quality)
- Source selection (web, discussions, academic papers)
- Multimedia search (images, videos)
- File uploads (PDFs, text files, images)
- Search history and discover feed
- Widget integrations (weather, calculations, stock widgets)
- Private AI search engine
- Research assistant with citations
- Knowledge discovery with local hosting
- Academic paper search
- Multimedia content discovery
- Language: TypeScript 98.7%
- Frontend: Next.js (TypeScript)
- Search Backend: SearxNG (privacy-focused search aggregator)
- Database: SQLite via Drizzle ORM
- AI Integration: Ollama, OpenAI, Anthropic Claude, Google Gemini, Groq
- Deployment: Docker
- Open-source and self-hosted
- Project renamed to Vane (repository: https://github.com/ItzCrazyKns/Vane)
- Latest release: v1.12.2 (April 10, 2026)
- GitHub: 33.7k+ stars, 3.7k+ forks, 46 contributors
- 33 total releases
- Default port: 3000
- Docker image:
itzcrazykns1337/vane:latest (or itzcrazykns1337/perplexica:latest for legacy)
¶ History and References
Any questions?
Feel free to contact us. Find all contact information on our contact page.