Vald
vald.vdaas.org
- Free plan
- Linux
- API
- Self-hosted
What Vald is
Vald is a free, open-source engine for distributed approximate nearest-neighbor search on dense vectors. It uses the NGT algorithm by default and supports NGT and Faiss index types, while allowing another vector search engine to be selected. Asynchronous automatic indexing can continue while search requests are served. Built around Kubernetes and cloud-native deployment, Vald supports horizontal scaling for memory and CPU, index replication, and automatic replica rebalancing when an agent goes down. Backups can use Persistent Volumes, S3-compatible object storage, or both, with restoration when an agent pod starts. Ingress filters can vectorize or preprocess data; egress filters can post-process results, using Vald’s gRPC interface. Client libraries are available for Go, Java, Python, and Node.js, with gRPC protobufs for other languages. Deployment documentation covers Kubernetes with Helm or its Helm operator, plus Docker agent deployment. Getting started requires Kubernetes, Helm v3 or later, and Go v1.15 or later.
Who it suits
Vald suits developers and infrastructure teams deploying self-hosted vector search on Kubernetes or Docker. It is intended for users comfortable managing clusters, Helm, and Go requirements.
Good to know, and what to watch
- Free open-source self-deployed software
- Asynchronous indexing during searches
- Horizontal Kubernetes scaling
- Index replication and rebalancing
- Go, Java, Python, and Node.js clients
- Requires a Kubernetes cluster
- Requires Helm v3 or later
- Requires Go v1.15 or later
- Filter components must implement Vald’s gRPC interface
Plans
Every plan and priceVald
Open source
The verdict
Vald provides distributed vector search with asynchronous indexing, replication, backups, and scaling controls. Its self-hosted model offers flexibility but requires Kubernetes-oriented deployment work.
What the maker publishes
- Backup
- Index backup can use Persistent Volumes, S3-compatible object storage, or both, and saved indexes can be restored when an agent pod starts. source
- Deployment
- The maker documents deployment on Kubernetes with Helm or its Helm operator, and also provides a tutorial for running a Vald agent on Docker. source
- Filtering
- Ingress filters can vectorize or preprocess data, while egress filters can post-process search results; filter components must implement Vald’s gRPC interface. source
- Indexing
- Vald supports asynchronous automatic indexing while continuing to serve search requests during indexing. source
- Product
- Vald is a highly scalable distributed approximate nearest neighbor engine for dense vector search. source
- Requirements
- The getting-started guide requires a Kubernetes cluster, Helm v3 or later, and Go v1.15 or later. source
- Resilience
- Vald supports index replication and automatically rebalances replicas when an agent goes down. source
- Scaling
- Vald is Kubernetes and cloud-native based, with horizontal scaling for memory and CPU. source
- SDKs and APIs
- Official client libraries are available for Go, Java, Python, and Node.js, and Vald publishes gRPC protobufs for other languages. source
- Search algorithm
- Vald uses the NGT approximate nearest neighbor algorithm by default and says users can switch to another vector search engine. source
- Security configuration
- Vald clusters can enable Kubernetes network policies to control ingress and egress traffic, subject to cluster prerequisites. source
- Support
- The project recommends its community Slack for questions and also accepts email at [email protected]. source
Checked against the maker's pages on 10 Oct 2026.