Local Deep Research

Local Deep Research is a self-hosted research assistant that performs multi-step, cited web research entirely on your own infrastructure. Its GitHub repository has gained more than 9,300 stars, and the project is in very active development with updates through October 2026.

Overview

Deep research features from big AI vendors are powerful but send your questions and sources to the cloud. Local Deep Research takes the opposite approach: it runs locally, pairs a language model with real search, and stores everything in an encrypted local database. The system plans a research strategy, queries more than ten search engines and scholarly sources, reads and cross-checks the results, and writes a structured report with citations. On the SimpleQA benchmark it reaches roughly 95 percent accuracy with a capable local model, showing that privacy does not have to mean weak answers.

Key Features

• Private by design: questions, sources and reports stay on your machine in an encrypted database
• Searches 10+ engines and academic sources including arXiv, PubMed and Semantic Scholar
• Can include your own private documents in the research corpus
• Works with local models via Ollama and llama.cpp, or cloud APIs from OpenAI, Anthropic and Google
• Generates structured, citation-backed reports rather than bare chat answers
• Self-hostable with Docker, including home server and homelab setups

💳 Pricing

Local Deep Research is free and open source under the MIT License. Run it with local models and it costs nothing at all; using cloud model APIs is optional and billed by those providers.

Who Is It For

This tool is for privacy-conscious researchers, journalists, analysts and homelab enthusiasts who want deep-research capability without outsourcing their curiosity to a vendor. It is also a strong option for organizations whose research topics are too sensitive for third-party cloud services.