What is Upstash Vector?

Upstash Vector is a serverless vector database designed for high-performance vector search at scale. It powers advanced systems in areas such as:

• Natural Language Processing (NLP)
• Retrieval Augmented Generation (RAG)
• Recommendation engines
• Image recognition
• Clustering or classification

… without any infrastructure management.

​

What is a Vector Database?

Vector databases are specifically designed to store and efficiently search high-dimensional vectors. These vectors can represent images, sounds, text, or other media for powerful similarity search across content that traditional databases struggle to represent properly.

Their specialization makes vector stores the perfect foundation for similarity-based search. Using specialized approximation algorithms such as Approximate Nearest Neighbor (ANN), they can typically provide much higher indexing and query performance than general-purpose databases or extensions like pgvector.

​

Upstash Vector Core Features:

• High-Performance Queries: Upstash Vector is powered by DiskANN[1], a highly efficient approximation algorithm that delivers very high recall rates (resulting in better output quality) and ultra-low latency. See ANN Search Algorithm for more insight into the technology behind our vector database.

• Batteries Included: A convenient REST API and SDKs with first-class Python and TypeScript support.

• Serverless Pricing: Upstash Vector is fully managed and serverless, we take care of the hosting and high availability complexities. You only pay for what you actually use. See our Vector Pricing Page for more details.

• Multiple Similarity Functions: We natively support similarity functions such as Euclidean distance, cosine similarity, and dot product. See Vector Similarity for more details.

• Metadata Support: Attach metadata to vectors to store additional information or context. When querying an index, you can then include a metadata filter to limit the search to records that match a filter expression. See Metadata Documentation for more details.

​

References

1. 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]