Vector databases · Vespa.ai (originally Yahoo)
Vespa
Open-source big-data serving engine combining vector search, keyword search and ranking at large scale.
Vespa is an open-source platform for low-latency serving of search, recommendation and ranking applications over large datasets, originally built at Yahoo and used internally for years before being open sourced. Unlike vector-only databases, Vespa combines dense vector (ANN/HNSW) search, traditional keyword search, structured filtering, and multi-phase machine-learned ranking within the same query, which makes it suited to applications that need relevance ranking beyond raw similarity scores, such as e-commerce search or content recommendation. It is a distributed system designed to scale horizontally across many nodes and handle both real-time writes and large-scale reads simultaneously. Vespa is self-hosted by default, typically on Kubernetes or bare infrastructure; Vespa Cloud is the managed hosting option run by the company behind the project. It requires more operational and modeling investment than simpler vector databases in exchange for more control over ranking behavior.
At a glance
| Vendor | Vespa.ai (originally Yahoo) |
|---|---|
| Pricing model | Free tier + paid plans |
| Free tier | Yes |
| Deployment | Cloud, Self-hosted |
| Open source | Yes (Apache-2.0) |
| Best for | Teams building search or recommendation products that need custom relevance ranking, not just similarity scores. |
Pricing
Self-hosted is free; Vespa Cloud offers a free trial tier with usage-based paid plans, though the pricing page did not publish specific dollar tiers.
Pricing has not been verified yet — see the vendor's site.
Features
- Combined vector, keyword and structured search in one query
- Multi-phase machine-learned ranking
- Horizontally scalable distributed architecture
- Real-time writes with large-scale concurrent reads
- Tensor computation for custom ranking models
- Managed Vespa Cloud option
Integrations
Profile last reviewed September 21, 2026