DeepSearcher
An open-source deep-research tool that combines large language models with vector databases to search, evaluate and reason over private data and produce a sourced report.
DeepSearcher combines large language models such as OpenAI's, Qwen, DeepSeek, Grok and Claude models with vector databases including Milvus and Zilliz Cloud to search, evaluate and reason over private data, aimed at enterprise knowledge management, question answering and information retrieval.
It supports partitioned data storage for efficient retrieval across supported vector databases, is compatible with multiple embedding models, and can integrate online content alongside private data when needed for a more accurate answer. A document loader currently supports local files, with web crawling under development.
It is written in Python, installs with pip or in an editable development setup with uv, and its quick-start demo requires an API key for the configured model provider.