ItzCrazyKns/Perplexica

Perplexica is a privacy-focused, open-source AI-powered answering engine designed to run entirely on your own hardware. Often described as an open-source alternative to Perplexity AI, Perplexica combines real-time internet search with the intelligence of large language models to deliver accurate, cited answers to complex queries without compromising user privacy. At its core, Perplexica leverages SearxNG as its search backbone, a privacy-respecting metasearch engine that aggregates results from multiple sources without tracking users. The retrieved results are then processed through a Retrieval-Augmented Generation (RAG) pipeline, where an LLM synthesizes the information into coherent, source-cited responses. This architecture ensures that every answer is grounded in verifiable web content rather than relying solely on the model's training data. One of Perplexica's most compelling features is its multi-model flexibility. Users can connect to virtually any LLM provider, including OpenAI, Anthropic Claude, Google Gemini, Groq, and locally hosted models through Ollama. This means developers and privacy-conscious users can run the entire stack on-premises with no data leaving their network, or mix cloud and local models depending on the task. Perplexica offers three distinct search modes tailored to different needs. Speed Mode prioritizes quick answers for simple lookups. Balanced Mode handles everyday research with a good tradeoff between depth and response time. Quality Mode performs deep, multi-step research for thorough investigation of complex topics. Beyond web search, the engine supports academic paper search, discussion forum search, image and video search, and domain-restricted queries. The platform also includes smart contextual widgets that surface relevant quick-lookup information such as weather forecasts, mathematical calculations, and stock prices directly in the search interface. Users can upload files including PDFs, text documents, and images for the AI to analyze alongside web results. All search history is saved locally, giving users full control over their data. With over 31,000 GitHub stars, 3,300 forks, and 44 contributors, Perplexica has established itself as one of the most popular open-source AI search projects. The project ships with Docker support for easy deployment, including a bundled SearxNG option that gets everything running with a single command. One-click deployment is also available through platforms like Sealos, RepoCloud, and Hostinger. A developer-facing API allows integration of Perplexica's search capabilities into custom applications and workflows.