This project provides a web application for conducting deep research using various Large Language Models (LLMs) and search engines. It's designed for users who need to quickly generate comprehensive research reports, offering features like multi-LLM support, web search integration, and local data storage for privacy.

How It Works

Deep Research leverages a "Thinking" and "Networking" model architecture to balance speed and depth in report generation. It integrates with multiple LLMs (Gemini, OpenAI, Anthropic, etc.) and search engines (Searxng, Tavily, Exa, etc.), allowing users to select their preferred tools. The application supports both local and server-side API calls, with all research data stored locally for enhanced privacy.

Quick Start & Requirements

• Local Development: pnpm install then pnpm dev (Node.js 18.18.0+ required).
• Docker: docker pull xiangfa/deep-research:latest then docker run -d --name deep-research -p 3333:3000 xiangfa/deep-research.
• Deployment: Supports Vercel and Cloudflare Pages.
• API Keys: Required for non-Gemini LLMs.
• Docs: https://github.com/u14app/deep-research

Highlighted Details

• Generates research reports in approximately 2 minutes.
• Supports a wide range of LLMs and web search tools.
• Offers local data storage and privacy-focused design.
• Built with Next.js 15 and Shadcn UI.

Maintenance & Community

• Active development indicated by release tags.
• Contributions are welcomed via pull requests.
• Contact via GitHub issues.

Licensing & Compatibility

• MIT License.
• Permits free use, modification, and distribution for commercial and non-commercial purposes.

Limitations & Caveats

The project notes that Docker image versions may lag behind the latest code by 1-2 days. Users running local Ollama or SearXNG instances might encounter CORS issues, recommending server proxy mode or configuring cross-domain requests for those services.