Vespa

Vespa is an AI search platform for building search, retrieval, recommendation and personalization applications. It brings together text search, a vector database, structured data filtering, tensor computation and machine learned model inference in a single engine, so retrieval and ranking happen together rather than across several systems. The vendor positions it for applications that need relevance at scale: retrieval augmented generation and AI agents that depend on accurate context, ecommerce and semi structured navigation, recommendation and ad targeting, and visual retrieval. It states that it handles billions of constantly changing data items and thousands of queries per second with latencies below 100 milliseconds, with real time updates. Hybrid search, multi vector representations and custom relevance models can be deployed without the limits of a pure vector similarity approach. Industry pages cover financial services, legal research, life sciences, media, telecommunications, travel, digital commerce and market intelligence, and the vendor cites Perplexity as a case study. Vespa is aimed at engineering teams building production search and AI applications. It can be used through Vespa Cloud, on AWS, or by building an application directly from the documentation, with a free trial available. Vespa Cloud pricing is usage based. The pricing page describes paying for what you use, with tiers for development, production and enterprise support, and $300 in free credits to start. Specific unit rates are not shown in the page text.