Upstash Vector is a serverless vector database designed for working with vector embeddings.
In the domain of databases, a vector database is essential for managing numeric representations of objects(images, sounds, text, etc.) in a multi-dimensional space. These databases are focused on efficiently handling vectors for storage, retrieval, and, most importantly, querying based on similarity. They are instrumental in integrating personalized data into AI applications enabling the AI system to provide tailored answers derived from your own dataset rather than generic responses, thereby enhancing the relevance and specificity of the generated insights.
Serverless Architecture and Pricing: Upstash Vector operates on a serverless model, abstracting away hosting and management complexities. Engineers are billed based on API calls, ensuring a transparent and cost-effective pricing structure aligned with our general philosophy and practices across all products. See pricing for more details.
Low-Cost High-Performance Queries: Improved upon to the DiskANN, Upstash Vector enables engineers to execute queries with high recall rates and low latencies compared to exhaustive search methods. DiskANN’s efficiency significantly enhances the performance of vector queries, making it a powerful tool for data retrieval and analysis. See Approximate Nearest Neighbor Search to get more insight on the tech behind the Upstash Vector.
Vector Similarity Functions: Powered by DiskANN, Upstash Vector supports different similarity functions, including Euclidean distance, Cosine similarity, and Dot Product. See Vector Similarity for more details.
REST API and SDKs: The product is equipped with a REST API and SDKs (Python and TypeScript) for seamless integration into engineering workflows. Explore the technical details of these interfaces to incorporate Upstash Vector into your codebase effortlessly.
Metadata Support: Upstash Vector allows users to attach metadata to vectors, enhancing data context. This feature contributes to referencing the original content upon retrieval and also filtering out the results to optimize the results further. See Metadata Feature for more details.
- Subramanya, S. J., Devvrit, Kadekodi, R., Krishaswamy, R., Simhadri, H. V. (2019). DiskANN: Fast Accurate Billion-Point Nearest Neighbor Search on a Single Node. In Proceedings of the 33rd International Conference on Neural Information Processing Systems (NeurIPS ‘19), Article No.: 1233, Pages 13766–13776. [https://dl.acm.org/doi/abs/10.5555/3454287.3455520]