Mixedbread AI integration for Haystack. This package provides seamless access to Mixedbread's multimodal AI capabilities, enabling intelligent search that understands meaning across text, images, code, PDFs, and diverse document types. Use our state of the art embedding and reranking models as part of your haystack workflows.
- MixedbreadTextEmbedder - State-of-the-art embedding models that generate vectors capturing deep contextual meaning for single texts and queries
- MixedbreadDocumentEmbedder - Embed full documents using advanced embedding models
- MixedbreadReranker - Powerful semantic reranking that significantly boosts search relevance
- MixedbreadDocumentParser - Layout-aware document parsing supporting PDF, PPTX, HTML and more formats
- MixedbreadVectorStoreRetriever - AI-native search engine that enables conversational queries across multimodal data
pip install mixedbread-ai-haystackGet your API key from the Mixedbread Platform and set it as an environment variable:
export MXBAI_API_KEY="your-api-key"from mixedbread_ai_haystack import MixedbreadTextEmbedder
embedder = MixedbreadTextEmbedder(model="mixedbread-ai/mxbai-embed-large-v1")
result = embedder.run(text="What is the capital of France?")
embedding = result["embedding"]All components support async operations:
import asyncio
async def embed_text():
embedder = MixedbreadTextEmbedder()
result = await embedder.run_async(text="Async embedding example")
return result["embedding"]
embedding = asyncio.run(embed_text())See the examples/ directory for complete usage examples:
- Embedders - Text and document embedding
- Reranker - Document reranking
- Document Parser - File parsing
- Vector Retriever - Vector-based search
# Install dev dependencies
pip install -e .[dev]
# Run all tests
python run_tests.py all
# Run only unit tests
python run_tests.py unit
# Run only integration tests (requires API key)
python run_tests.py integration
# Run specific test files
python run_tests.py tests/test_text_embedder.pyLearn more at mixedbread.com/docs:
Apache 2.0 License