RAG et connaissances
llmware
Unified framework for enterprise RAG pipelines with small, specialized models - local, private and secure.
★ 14,8 k étoilesApache-2.0
Voir le guide →UltraRAG est un projet de RAG et connaissances open source, créé par OpenBMB (OpenBMB/UltraRAG sur GitHub). Il compte 5 717 étoiles sur GitHub et 450 forks, il est publié sous la licence Apache-2.0 et a été mis à jour pour la dernière fois sur GitHub le 7 octobre 2026.
A low-code MCP framework for building complex, innovative RAG pipelines — visibly, not as a black box.
Map the retrieval and generation steps you need — corpus processing, retriever, reranking, generation, evaluation.
Clone the repo and sync dependencies with uv, choosing core, full or per-module extras.
Point the pipeline at your embedding and LLM providers (for example OpenAI, Qwen or local vLLM) and prepare your corpus.
Execute a bundled example experiment YAML, then open the UI to inspect each step's output.
Compose retrieval, generation and evaluation steps in a few dozen lines of YAML instead of hand-writing orchestration code.
Built as a low-code MCP framework, so pipeline components interact through the Model Context Protocol.
Designed against black-box development: pipeline logic and intermediate results stay inspectable, including in the web UI.
Install only the modules you need with extras — retriever, generation, or the full stack with corpus processing and evaluation.
Ships multimodal support and example pipelines including a lightweight DeepResearch setup.
Ready Docker images (base CPU, base GPU and full GPU) are published on Docker Hub for a fast start.
Clone UltraRAG (shallow) and enter the directory.
git clone https://github.com/OpenBMB/UltraRAG.git --depth 1
cd UltraRAGCreate the environment with uv; add --all-extras for the full feature set.
uv syncExecute the hello-world experiment to verify the installation end to end.
ultrarag run examples/experiments/sayhello.yamlUltraRAG is maintained by OpenBMB, the team behind MiniCPM and related open models.
Yes — it is open source under the Apache-2.0 license.
It depends on your pipeline: hosted models (OpenAI and similar) need your key, while local models served with vLLM do not.
Those are general application frameworks; UltraRAG specializes in declarative, MCP-based RAG pipelines where every reasoning step remains visible and swappable.
Clone the repo, run uv sync, and execute the sayhello example pipeline.
Unified framework for enterprise RAG pipelines with small, specialized models - local, private and secure.
RAG for your monorepo: query, understand and edit multi-language codebases with a knowledge graph.
Free rag & knowledge tool that enterprise-ready vector database toolkit for building s.
Self-hosted rag & knowledge: cometchat docs search + implementation bundles: add cha.
Complete API layer for private AI apps on local models: RAG, skills, tools, MCP and text-to-SQL
MinerU — turns PDFs and Office documents into LLM-ready Markdown and JSON for agents and RAG pipelines, Apache-2.0 plus terms
Les autres projets de la catégorie RAG et connaissances avec le plus d'étoiles du catalogue LibreAI — la sélection réaliste d'alternatives si UltraRAG ne vous convient pas.
Build Agentic workflows, RAG pipelines, with rich AI model and tool support on one collaborative workspace. Deploy on cloud, VPC, or…
RAGFlow is a leading open-source Retrieval-Augmented Generation (RAG) engine that fuses cutting-edge RAG with Agent capabilities to create a superior context…
Lightweight OCR toolkit turning PDFs and images into structured, LLM-ready data in 100+ languages
MinerU — turns PDFs and Office documents into LLM-ready Markdown and JSON for agents and RAG pipelines, Apache-2.0 plus terms
Stop renting your intelligence. Own it with AnythingLLM. Everything you need for a powerful local-first agent experience
Projets de la catégorie RAG et connaissances proches dans le catalogue, les plus proches en étoiles de UltraRAG — continuez à explorer de part et d'autre.
An OLTP graph database with native vector and full-text search, built in Rust on object storage.
Build Neo4j knowledge graphs from unstructured data using LLMs — then chat with your graph.
Build a ChatGPT over your own data using natural language and LlamaIndex.
RAG for your monorepo: query, understand and edit multi-language codebases with a knowledge graph.
SQLBot — DataEase's conversational analytics: ask in natural language, get SQL answers via LLMs plus RAG.
An agent that finds anything on your computer — federated search over your files, in place.
Gratuit et ouvert : récupérez le code source, mettez une étoile au dépôt et commencez à créer.