Perplexica: An Open-Source AI-Powered Search Engine 131

Perplexica is an open-source AI-driven search engine that offers multiple search modes, providing users with a more precise and intelligent search experience. This search engine is inspired by Perplexity AI and can adjust its search algorithm based on user needs to obtain more relevant search results. At the same time, Perplexica uses advanced machine learning algorithms to optimize its search capabilities.

Perplexica supports locally deployed large language models, such as Llama3 and Mixtral, and offers multiple search modes, including Copilot mode, Normal mode, and Focus mode, with the latter providing precise answers for specific types of questions. At the same time, Perplexica uses SearxNG as a meta-search engine to ensure the latest information is retrieved, and supports image and video search. Future plans include introducing more Focus modes and refining the Copilot mode.

Perplexica Features:

• Uses local large language models (LLMs) such as Llama3 and Mixtral.

• Two main modes: Copilot mode (in development) and Normal mode.

• Six Focus modes, including All mode, Writing Assistant mode, Academic Search mode, YouTube Search mode, Wolfram Alpha Search mode, and Reddit Search mode, which adjust the search algorithm according to user needs to obtain more relevant search results.

• Always uses SearxNG to obtain the latest information, without the need for daily data updates.

• Supports image and video search.

• Uses advanced machine learning algorithms, such as similarity search and embedding techniques, to refine search results.

• Ensures privacy protection, using SearxNG as a measure for ensuring timeliness, avoiding the overhead of daily data updates.

Planned features include completing Copilot mode, adding a settings page, supporting local LLMs, adding discovery and history saving features, and introducing various Focus modes.

Perplexica leverages advanced machine learning algorithms to optimize search results and emphasizes privacy protection. This open-source project is committed to becoming a comprehensive and efficient search solution that meets users' needs for various types of information while protecting user privacy and providing the latest search results.