farm-haystack

What it is and what it does

Haystack is a framework for building NLP applications powered by large language models, transformers, and vector search. It provides a pipeline architecture where data flows through nodes—each performing a single task like document retrieval, preprocessing, or LLM inference—and an agent system that uses an LLM to decide the next action dynamically. The framework includes built-in support for multiple document stores (Elasticsearch, Weaviate, Pinecone, FAISS), language models from OpenAI and Hugging Face, and tools for semantic search, question answering, and retrieval-augmented generation.

The package depends on 23 runtime libraries including pandas, scikit-learn, transformers, tiktoken, and others for data handling, ML operations, and LLM integration. However, Haystack 1.x reached end-of-life on March 11, 2025 and is no longer maintained. The maintainers have released Haystack 2.x as a separate package (haystack-ai) with improved pipeline flexibility, better component composition, and production-ready features like structured logging and tracing. Existing users are strongly encouraged to migrate.

Use it for:

• Build retrieval-augmented generation (RAG) systems that combine vector search with LLM generation over custom documents.
• Create question-answering applications that find and synthesize answers from large document collections in natural language.
• Perform semantic document search and ranking based on meaning rather than keyword matching.
• Develop agent-based systems that use an LLM to reason over multiple tools and data sources to resolve complex queries.
• Fine-tune transformer models on domain-specific data and integrate them into end-to-end NLP pipelines.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Haystack is an end-to-end NLP framework for building LLM applications with pipelines, agents, and retrieval-augmented generation (RAG), connecting components like vector databases, language models, and document stores.

No—do not install this package for new projects. Haystack 1.x reached end-of-life on March 11, 2025 and receives no further updates or support. The final version is 1.26.4. Migrate to Haystack 2.x (distributed as haystack-ai) instead, which has been stable since March 2024 and includes more flexible pipelines, better component customization, improved integrations, and production-ready features. Existing 1.x codebases may continue to run, but new development should use haystack-ai.

Install

farm-haystack on PyPI

pip install farm-haystack

uv add farm-haystack

poetry add farm-haystack

Installing farm-haystack

Before you install

Low install friction with a pure-Python wheel. The package is actively maintained with recent commits and a large community (26210 stars), though it reached end-of-life on March 11, 2025 and is no longer receiving updates—the maintainers recommend migrating to Haystack 2.x via the separate haystack-ai package.

License in practice

Apache-2.0 permissive license allows commercial use, modification, and distribution with minimal restrictions, suitable for most production deployments.

Quickstart

pip install farm-haystack

from haystack.pipelines import Pipeline
from haystack.nodes import Retriever, PromptNode

pipeline = Pipeline()
pipeline.add_node(Retriever(), name="retriever", inputs=["Query"])
pipeline.add_node(PromptNode(), name="prompt", inputs=["retriever"])

Requires Python 3.8 or later. The package reached end-of-life on March 11, 2025 and is no longer receiving updates; new projects should use haystack-ai instead.

Verify before relying

• Exact performance characteristics and scalability limits for document retrieval at production scale.
• Compatibility and testing status with specific vector database versions (Weaviate, Pinecone, FAISS, etc.).
• Migration effort and breaking changes when upgrading from 1.x to Haystack 2.x (haystack-ai).

Package facts

Evidence: farm_haystack-1.26.4.post0-py3-none-any.whl

Keywords: BERT, QA, Question-Answering, Reader, Retriever, albert, language-model, mrc, roberta, search, semantic-search, squad, transfer-learning, transformer

Tags

More Artificial Intelligence packages

Further reading