Open Source
Other RAGPipelinePythonProductionComponentsdeepset
Haystack
Deepset's open-source framework for building production-ready LLM pipelines. Component-based architecture for RAG, question answering, summarisation, and agent systems.
Install
$pip install haystack-ai About
Haystack 2.0 is a complete rewrite with a component-based pipeline architecture. Any Python class decorated with @component becomes a pipeline node with typed inputs and outputs that Haystack validates at connection time.
It is production-oriented: pipelines serialise to YAML for version control, the evaluation framework measures retrieval recall and answer faithfulness, and deepset Cloud provides a managed deployment environment.
Getting Started
pip install haystack-ai
from haystack import Pipeline
from haystack.components.generators import AnthropicGenerator
from haystack.components.builders import PromptBuilder
pipeline = Pipeline()
pipeline.add_component("prompt", PromptBuilder(template="Answer: {{query}}"))
pipeline.add_component("llm", AnthropicGenerator(model="claude-sonnet-4-5"))
pipeline.connect("prompt", "llm")
result = pipeline.run({"prompt": {"query": "What is RAG?"}})
print(result["llm"]["replies"][0])Features
Pipeline API — DAG of connected components for LLM workflows
50+ integrations: OpenAI, Anthropic, Cohere, HuggingFace models
Document stores: Elasticsearch, Weaviate, Chroma, Qdrant, pgvector
Advanced retrievers: BM25, dense, hybrid
Haystack Agent — tool-using agent with memory
Evaluation framework with RAGAS and DeepEval support
Serialisable pipelines — save to YAML and reload
deepset Cloud — managed deployment platform
References & Links
Project Info
Language Python
License Apache-2.0
Stars 21.0k ★
Forks 2.3k
Category Other
Topics
RAGPipelinePythonProductionComponentsdeepset
Ready to try Haystack?
Free and open source. Install in one command, works with Claude Code, Cursor, Windsurf, and more.
$pip install haystack-ai