#5555th of 60 in RAG, knowledge and searchFor science, health and research, research and papers
deep-research
Next.js app that runs multi-step web research and writes a Markdown report
- Stars
- 4.7k
- License
- MIT
- Last commit
- Jun 2026
- Last release
- Feb 2026
- Language
- JavaScript
4,448 on 22 Apr 20264,685 on 11 Nov 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
Deep Research takes a topic, asks clarifying questions, drafts a research plan, then runs rounds of search queries and compiles a Markdown report using separate thinking and task models. It works with many LLM providers (Gemini, OpenAI, Anthropic, Ollama, OpenRouter and others) and search backends (Searxng, Tavily, Firecrawl, Exa, Brave). It also exposes an SSE API and an MCP server.
Who it is for: Self-hosters who want deep-research reports using their own LLM and search keys
Strengths
- Supports many LLM providers, including Ollama and any OpenAI-compatible endpoint
- Pluggable search: Searxng, Tavily, Firecrawl, Exa, Bocha, or the model's own search
- Exposes /api/sse and an MCP server (streamable-http and SSE) for use from other tools
- Research history and local knowledge base from uploaded files, stored in the browser in local mode
Weaknesses
- Calling Ollama or Searxng from the browser can hit CORS errors; server proxy mode needed
- Custom model list works only in proxy mode
- MCP clients need long timeouts (README example uses 600) because runs are slow
- Docker image can lag the latest release by one to two days
What it needs
- no GPU
- Docker + Compose
- Needs LLM API key or Ollama, Search provider (Searxng, Tavily, Firecrawl, Exa, Brave, or model search), Node.js 18.18+ for development
- Models: Gemini, OpenAI, Anthropic, DeepSeek, Grok
- port 3000
What changed
From the release notes. No flag means the note does not say so, not that the upgrade is safe.
Also in RAG, knowledge and search
See all 60| Rank | Project | Score |
|---|---|---|
| 1 | RAGFlowRAG engine with deep document parsing, template chunking and agentic retrieval | 90 out of 100 |
| 2 | Open NotebookSelf-hosted NotebookLM alternative for research notebooks, chat, and podcasts | 90 out of 100 |
| 3 | LightRAGGraph-based RAG framework with a REST server and WebUI | 90 out of 100 |
| #4 | deeptutorTutoring workspace with knowledge bases, guided learning, memory and quizzes | 89 out of 100 |
| #5 | MaxKBEnterprise knowledge-base agent platform with RAG, workflows and MCP tools | 89 out of 100 |