Upstash Vector

What Upstash Vector is

Upstash Vector is a cloud-based, serverless vector database for applications that need vector search, including natural language processing, retrieval-augmented generation, recommendations, image recognition, clustering, and classification. It uses DiskANN for approximate nearest-neighbor queries; Upstash says this provides high recall and low latency. FreshDiskANN is also listed as an index type. Developers can connect through a REST API or SDKs, with first-class support for Python and TypeScript and additional SDKs for Go and PHP.

The database supports Euclidean distance, cosine similarity, and dot product. Vectors can carry metadata, and query filters can select results using metadata expressions. Hybrid indexes combine dense semantic search and sparse exact-token search, then fuse their results. Documented integrations include Vercel AI SDK, Flowise, LangChain, Langflow, LlamaIndex, and LlamaParse. The default topk limit is 1,000; each vector can have up to 48 KB of metadata and 1 MB of data. The Free plan allows 10K daily queries or updates, 1,536 maximum dimensions, 100 namespaces, and 1 GB of data and metadata. Pay as You Go costs $0.40 per month, billed per 100K requests, with storage at $0.25/GB and bandwidth free up to 200 GB/month, then $0.03/GB.

Who it suits

Upstash Vector suits developers adding semantic or hybrid vector search to cloud applications, especially those using Python or TypeScript. Its REST API, SDKs, metadata filters, and listed integrations support a range of search and machine learning use cases.

Limits

The Free plan allows 10K daily queries or updates, 1,536 maximum dimensions, 100 namespaces, and up to 1 GB of data and metadata. It also has a default topk limit of 1,000, and each vector is limited to 48 KB of metadata and 1 MB of data.

Getting started

1. Open the Upstash Vector documentation website.
2. Choose the Free, Pay as You Go, Fixed, or Enterprise plan.
3. Connect through the REST API or an SDK for Python, TypeScript, Go, or PHP.
4. Create an index and use a supported similarity function for queries.
5. Add metadata and query filters, or use a hybrid index for dense and sparse search.

Verdict

Choose Upstash Vector if you need a serverless database for vector search with metadata filtering, hybrid indexes, and developer integrations. The free tier has daily query/update and data limits; teams needing higher capacity, more dimensions, enterprise security, or dedicated support can consider the other plans, with Enterprise pricing available by contacting Upstash.