Free
5 GB storage; 2.5 million vCUs per month.
AI Search & Research Tools

PRODUCT OVERVIEW
Zilliz Cloud is a fully managed Vector Lakebase service built by the creators of Milvus. It unifies real-time vector serving, iterative discovery, and batch analytics on a single source of truth, scaling to hundred-billion-entity datasets. Full-spectrum search combines vector, full-text, JSON, and geospatial retrieval with hybrid filtering and reranking. Tiered architectures and on-demand compute let teams balance latency, throughput, and cost, while lake-native storage and zero-copy access to external data support large-scale AI workloads. Intended for developers and enterprises building AI applications such as retrieval-augmented generation, semantic search, and recommendation systems, it offers a free tier, serverless and dedicated clusters, enterprise controls, and BYOC deployment.
Sign up for Zilliz Cloud from the official website to access the console and free tier.
Select a free, serverless, or dedicated cluster based on your workload's latency, throughput, and scale requirements.
Use the core APIs or CLI to create collections and load vector and scalar data into your cluster.
Execute vector, full-text, hybrid, and filtered searches, and use on-demand compute for lake-scale indexing and analytics jobs.
Building retrieval-augmented generation (RAG) applications
Semantic search and recommendation systems
Large-scale AI data operations and analytics
Multi-tenant AI applications with unlimited namespaces
Real-time serving for high-QPS, low-latency workloads
Cost-sensitive batch analytics on massive vector datasets
A compute unit (CU) is a group of hardware resources for serving your indexes and search requests. It can be considered a fully-managed physical node for deploying search service.
A virtual compute unit (vCU) measures resources consumed by read operations (search, query) and write operations (insert, upsert, delete). Read and write costs vary for different vCU usage.
Choose Performance-optimized for instant search and high concurrency, Capacity-optimized for large datasets with reliable speeds, and Tiered-storage for ultra-large-scale, cost-sensitive workloads.
Performance-optimized supports up to 2 million 768-dimensional vectors per CU, Capacity-optimized up to 8 million, and Tiered-storage up to 40 million. Additional scalar fields may reduce capacity.
Yes, the Free plan includes 5 GB storage, 2.5M vCUs per month, and up to 5 collections, intended for learning and personal projects.
The Enterprise plan includes a 99.95% uptime SLA, audit logs, SSO, granular RBAC, multi-replica and elastic scaling, private endpoint and VPC peering, and enterprise support.
Yes, BYOC (Bring Your Own Cloud) lets you deploy on your infrastructure of choice with the same features and experience as SaaS Dedicated clusters.
On-demand compute charges only for active job runtime, with Query CU and Indexing CU priced at $0.41 per CU-hour, billed by CU-minute.
Official pricing checked 2026-11-09. Dedicated prices are starting prices; usage, storage and deployment configuration affect the final bill. Standard Serverless starts at $0 and is usage-based.
5 GB storage; 2.5 million vCUs per month.
COMMUNITY REVIEWS
No reviews yet.