Zilliz Cloud Vector Lakebase

Zilliz is the creator of Milvus, the widely adopted open-source vector database, and is headquartered in Redwood Shores, California. It serves companies building retrieval-augmented generation, semantic search, recommendation and AI agent applications that need to store and search billions of embeddings reliably and at low cost.

Zilliz Cloud is the fully managed service built on Milvus, now extended into a Vector Lakebase. It offers performance-optimized, capacity-optimized and tiered-storage serving, hybrid search across dense and sparse vectors, text, JSON and geospatial data with reranking, on-demand pay-per-query search, and zero-copy search over Iceberg, Lance, Parquet and Vortex tables in object storage. It runs as serverless, dedicated or bring-your-own-cloud in more than 30 regions on AWS, Google Cloud and Azure. More than 10,000 enterprises use it, including NVIDIA, Salesforce, Reddit, Roblox, MiniMax and Sarvam.

• Serving tiers: performance-optimized, capacity-optimized and tiered-storage clusters.
• Hybrid search: dense and sparse vectors, full text, JSON and geospatial filters with reranking.
• On-demand search: pay-per-query compute over billion-scale vector data.
• Data lake search: zero-copy search over Iceberg, Lance, Parquet and Vortex tables.
• Global Cluster: cross-region replication and failover for high availability.
• Deployment choice: serverless, dedicated or bring-your-own-cloud on AWS, Google Cloud and Azure.

Vietnamese banks and insurers can power internal knowledge assistants over policy documents with hybrid vector search. E-commerce platforms can run semantic product search and recommendations at scale. Media and legal firms can search large archives of text and images for AI agents.

Production teams get single-digit-millisecond vector search at 1,000+ QPS with a 99.99% uptime SLA without running Milvus themselves. Tiered storage on S3 with hot cache can cut costs by up to 90%, and on-demand search cost about 1/15 of serverless in Zilliz's billion-vector benchmark. BYOC keeps data inside the customer's VPC. New work-email signups receive 100 USD in free credits.

Retrieval-augmented generation and AI agents need fast access to large volumes of unstructured data, making vector search core AI infrastructure. Enterprises in Southeast Asia building chatbots, search and recommendation in local languages increasingly need managed vector stores that scale from prototype to billions of records.

Headquartered in Redwood Shores, California; founder and CEO Charles Xie. Investors include Prosperity 7 Ventures, Pavilion Capital (Temasek), Hillhouse Capital, 5Y Capital, Yunqi Partners and Trustbridge Partners. Milvus has 44,000+ GitHub stars and over 100 million Docker pulls. Announced Milvus 3.0 in July 2026.