QAnything
NetEase Youdao's open-source local RAG question-answering system.
About
QAnything is an open-source, locally deployable knowledge-base question-answering system from NetEase Youdao. Developers deploy it on their own infrastructure to index files and web links, then ask questions over their documents with retrieval-augmented generation, keeping data fully private.
It is built on Youdao's BCEmbedding models, bilingual embedding and reranker models whose combination claims state-of-the-art results on RAG evaluations in LlamaIndex. The open-source LLM side is based on Qwen, and the project has attracted around 14,000 GitHub stars with active maintenance.
What it can do
- Local self-hosted RAG over files and web links
- BCEmbedding bilingual embedding and reranker models
- Cross-lingual Q&A, information extraction, and API access
- Docker Compose deployment with no data leaving your network
Setup
Clone netease-youdao/qanything and deploy with Docker Compose; GPU with 16GB+ VRAM recommended.
Prerequisites
Docker, NVIDIA GPU with 16GB+ memory, CUDA 12+, and docker compose.
Use cases
Works with
Trust score — 7/10
Scored 7/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 95; Trending score 72/100 from researcher assessment.
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.