stanford-oval/WikiChat

WikiChat is an improved RAG. It stops the hallucination of large language models by retrieving data from a corpus.

56
/ 100
Established

Implements a 7-stage LLM pipeline with structured information retrieval from Wikipedia (and custom corpora), using vector embeddings and LLM-based reranking to ground responses in retrieved passages with inline citations. Supports 25+ multilingual Wikipedias via a free rate-limited API, or can build custom indexes using Qdrant for vector search and Snowflake's Arctic embedding model. Integrates with 100+ LLMs through LiteLLM and LangChain, with optional multi-user deployment via Chainlit and Cosmos DB.

1,555 stars.

No Package No Dependents
Maintenance 10 / 25
Adoption 10 / 25
Maturity 16 / 25
Community 20 / 25

How are scores calculated?

Stars

1,555

Forks

140

Language

Python

License

Apache-2.0

Last pushed

Jan 31, 2026

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/rag/stanford-oval/WikiChat"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.