Deep Research

Deep Research is an open source web application that uses large language models to produce in depth research reports in a few minutes. A user enters a topic, the app asks clarifying questions, drafts a research plan, collects information from the web and local files, and writes a final report, moving through five visible stages that can each be refined and rerun. It suits researchers, analysts and developers who want a self hosted research assistant that works with the model provider of their choice. It supports Gemini, OpenAI, Anthropic, DeepSeek, Grok, Mistral, Azure OpenAI, OpenRouter, Ollama and any OpenAI compatible endpoint, and it can search through SearXNG, Tavily, Firecrawl, Exa, Brave and other search services. Separate thinking and task models balance depth against speed. A local knowledge base accepts text, Office and PDF files, reports can be edited in rich text or Markdown with adjustable reading level, length and translation, and a knowledge graph can be generated from the content. Research history is kept in the browser, and the app can also run as a service through an SSE API or plug into other assistants as an MCP server. It installs as a progressive web app and is available in English, Simplified Chinese and Spanish. The software is free under the MIT License. There is no paid plan; users pay only for the model and search APIs they connect, and the project recommends starting with a free Gemini API key. It can be deployed to Vercel or Cloudflare or run locally with Node.js. The hosted page at the listed address is a public instance of the app rather than a commercial service.