The company claims that the combination of sparse retrieval, dense retrieval, and reranking capabilities within its database can improve performance by up to 48 percent. The new proprietary reranking and embedding models, along with third-party models like Cohere’s Rerank 3.5, provide customers with quick and easy access to high-quality retrieval. Pinecone provides these capabilities via a single API, allowing developers to create GenAI retrieval applications without the burden of managing model hosting, integration, or infrastructure.
Key takeaways:
- Pinecone has updated its vector database to include fully managed embedding and reranking models, improving the performance and accuracy of AI-powered solutions.
- The database now includes new features such as a proprietary reranking model, a proprietary sparse embedding model, a new sparse vector index type, and integration of Cohere’s Rerank 3.5 model.
- Pinecone claims that by combining sparse retrieval, dense retrieval, and reranking capabilities within its database, developers can create retrieval systems that deliver up to 48 percent better performance than dense or sparse retrieval alone.
- Developers can now develop GenAI retrieval applications without the burden of managing model hosting, integration, or infrastructure, and customers can access Pinecone through the AWS Marketplace.