QuivrHQ/quivr

Project overview

quivr-core is an opinionated Python library that helps developers build Retrieval-Augmented Generation (RAG) systems quickly, supporting multiple LLM providers and file types. Adopt this project to reduce RAG design overhead and integrate document-based question answering into existing Python workflows.

Project type

RAG

Use cases

Knowledge Q&A

Deployment

Refer to project documentation

License

License pending

Best for

• Python developers who need an opinionated, code-first library to quickly implement document-based question answering.

Key capabilities

• Enables the creation of a 'Brain' instance from various files (such as PDF or TXT) to answer questions using Retrieval-Augmented Generation.
• Supports APIs from Anthropic, OpenAI, and Mistral, alongside local models running through Ollama.
• Allows developers to define custom RAG workflows using YAML configuration files for steps like filtering history, rewriting, retrieving, and generating.
• Provides configurable reranking of retrieved document chunks using external suppliers like Cohere.

Limitations and risks

• The project is at an early version (0.0.33), indicating that the API may change and features are still being added.

Getting started

• Users can install the package via pip in approximately 30 seconds and run a 5-line Python example script to achieve initial success.

Evidence and sources

• Release: core: v0.0.33
• README: - **Opiniated RAG**: We created a RAG that is opinionated, fast and efficient so you can focus on your product
• README: We take care of the RAG so you can focus on your product.
• README: - **LLMs**: Quivr works with any LLM, you can use it with OpenAI, Anthropic, Mistral, Gemma, etc.
• README: pip install quivr-core # Check that the installation worked

Reviewed at: 2026-07-17T08:16:07Z