Zilliz Cloud Pricing

Scalable pricing designed for every team to fit any budget. Estimate Cost

Subscribe on Marketplace.

Amazon Web Services

Google Cloud Platform

Microsoft Azure

Free
A starting point for learning, experimenting, and prototyping, with easy migration to paid plans.
$0/mo.
Quick Start
1. 5 GB storage
  Enough for 1M 768 dim vectors.
2. 2.5M
  vCUs
  A virtual compute unit (vCU) is used to measure the resources consumed by read operations (such as search and query) and write operations (such as insert, upsert, and delete). The read and write costs vary for different vCU usage.
  per month included
3. Up to 5 collections
Serverless
For applications with variable or infrequent traffic. Minimal configuration required.
$4 / Million
vCUs
A virtual compute unit (vCU) is used to measure the resources consumed by read operations (such as search and query) and write operations (such as insert, upsert, and delete). The read and write costs vary for different vCU usage.
Try Free
1. Pay only for what you use
2. Auto-scaling
3. Up to 100 collections
Dedicated
Dedicated clusters offer use case optimized CUs to achieve high control, consistent performance, and cost-effectiveness. Suitable for development and testing.
From
$99 /mo. (Up to 30-day free trial)
Try Free
1. Multiple cloud providers and regions
2. Use case optimized CU types
3. Basic metrics and monitors
BYOC
Designed for organizations prioritizing custom infrastructure, enhanced data protection, and compliance.
1. Deploy on your infra of choice
2. Enhanced data control and security
3. Flexibility and scalability on demand
Contact Us

Dedicated clusters offer diverse CUs tailored to fit your use cases

A CU (compute unit) is a measure of computational resources for data processing. Each CU type offers different combination of CPU, memory, and storage.

CU Type	Search QPS	Search Latency	Per CU Capacity	Cost per Million Vectors	Best Suited for
Performance-optimized	500~1500	sub-10 ms	1.5 million vectors	from $65/mo.	Ideal for real-time applications requiring instant search results and high concurrent traffic.
Capacity-optimized	100~300	tens-ms	5 million vectors	from $20/mo.	Perfect for applications handling large vector datasets while maintaining reliable search speeds.
New!Extended-capacity Contact Sales	5~20	hundreds-ms	20 million vectors	from $10/mo.	Designed for massive-scale datasets where optimizing total cost is prioritized over latency.

*This table is based on evaluations of 768-dimensional vectors.

View Plan Feature Details

Free
Quick Start
Serverless
Try Free
Dedicated (Standard)
Try Free
Dedicated (Enterprise)
Try Free

Deployment
Environment	Shared	Shared	Dedicated	Dedicated
CU type
Performance-optimized Ideal for applications requiring low latency and high throughput. Each CU can handle about 1.5 million 768-dim vectors.
Capacity-optimized Suited for managing large datasets with moderate search performance requirements. Each CU can handle about 5 million 768-dim vectors.
New!Extended-capacity Best for scenarios with extensive datasets where cost-efficiency is prioritized over latency. Each CU can handle about 20 million 768-dim vectors. Contact Sales
Public cloud provider	Google Cloud	Google Cloud	AWS, Google Cloud, Azure	AWS, Google Cloud, Azure
Scale Scale up and down with zero downtime.		Auto-scaling	Manual scaling to 32 CUs	Auto-scaling Manual scaling to 256 CUs or more
Uptime SLA Guaranteed uptime for production workloads.				99.95%

High Availability
Availability zone		Single	Single	Multiple
Replica Zilliz Cloud supports cluster-level replication, delivering a QPS that scales proportionally with the number of replicas. This replication feature automatically distributes replicas across different availability zones (AZs), enhancing both throughput and high availability.

Data Management
Cross-tier data migration Easy migration from Free tier, Serverless, and Standard.	From Free tier	From Free tier and Serverless	From Free tier, Serverless & Dedicated(Standard)
Migration from external sources Easy migration from Milvus, Pinecone, Qdrant, Elasticsearch, PostgreSQL, Tencent Cloud VectorDB, etc.
High speed data import High speed data import from object storage like S3.
Recycle bin Dropped collections will be retained for 30 days to facilitate easy recovery.

Data Security & Compliance
OAuth 2.0 OAuth 2.0 for authorizing account access without sharing or storing user login credentials.
Enterprise SSO Streamlined user authentication which supports both Okta and SAML 2.0 protocol.				Public Preview
MFA
Auditing Comprehensive auditing logs that capture all UI and RESTful API operations on the control plane, as well as all SDK and RESTful API operations on the data plane.
API key management
Data encryption in transit and at rest
Backup and restore Supports backup at both the cluster and collection levels, with options for manual and automatic backups.
IP address access control
Private networking Private connection between your VPC and Zilliz Cloud VPC.
SOC 2 Type II and ISO/ICE 27001 compliant, GDPR and HIPPA ready

Observability
Fine-grained metrics with real-time monitoring dashboards Metrics to monitor performance, storage, usage, data statistics, etc.
Alerts Seamless integration with various alerting channels including emails, PagerDuty, Slack, Opsgenie, Lark, Webhook, etc.
Alerting and monitoring integrations Monitoring API and integrations with Prometheus and Datadog.
Job Center A centralized job center page to track the progress of tasks including migration, data import, backup and restore, clone collection, and create sample collection, etc.

Role-based Access Control
Organization and project management	1 organization 1 project	1 organization Up to 10 projects	1 organization Up to 10 projects	1 organization Up to 10 projects
Organization and project RBAC Role-based access control at both organization and project levels.
Data plane RBAC Data layer RBAC enables precise permission control over collections, partitions, and operations, enhancing security and operational alignment.

Integrations and Tools
Intuitive RESTful APIs for control and data plane operations
User-friendly SDKs in multiple programming languages	Python, Java, Go, and Node.js SDKs	Python, Java, Go, and Node.js SDKs	Python, Java, Go, and Node.js SDKs	Python, Java, Go, and Node.js SDKs
VectorDBBench An open-source benchmarking tool for mainstream vector databases. It is also a tool for ultimate performance and cost-effectiveness comparison.

Support
Community support
Email support
Response time SLAs
Urgent	4 hours	4 hours	1 hour
High	1 business day	1 business day	4 hours
Normal	2 business days	2 business days	1 business day
Technical contacts	Up to 1	Up to 1	Up to 4

Pipelines
Data source connectors Both batch and streaming data source are supported.
Ingestion pipelines Ingestion pipelines transform your unstructured data to a searchable vector collection, streamlining doc parsing, chunking, embedding, vector indexing processes.
Search pipelines Search on text and image and supports advanced retrieval strategies such as dense and sparse embedding, multi-stage retrieval and reranking.

Pipelines

Streamline embedding, vector ingestion, search and reranking in one API service. Pipelines support data types including file, text and image, with open-source and 3rd party models. Skip the hassle of DevOps while enjoying the pay-as-you-go pricing.

Estimate Your Cost

Use this calculator to understand how Zilliz Cloud charges.

Cloud Provider

Cloud Region

Pricing Plan

CU Type

Number of Entities

An entity is a row of data. For example, an entity can represent an image or a doc chunk.

Million

Vector Dimensions

Vector dimension refers to the length of the vector and can vary depending on the embedding model you use.

Estimated total cost per month

Total Cost (excl. tax) = CU Cost × Replica Count + Storage Cost

The price estimate is monthly, but the actual cost is billed hourly. You can suspend clusters anytime to save costs.

CU Cost
CU
CU Usage
CU
Unit Price
$0/CU
A CU is the basic unit of compute resources used for parallel processing of data.
$
Storage Cost
GB
Storage Usage
GB
Unit Price
$0/GB
$

Try Free Contact us

Acknowledgement: Prices are estimates only and are not intended as actual price quotes. Actual pricing may vary depending on a variety of factors, including your actual usage of services.

Frequently Asked Questions

What is a Compute Unit (CU)?
A compute unit (CU) is a group of hardware resources for serving your indexes and search requests. You can simply consider a CU as a fully-managed physical node for deploying search service.
What is a vCU?
A virtual compute unit (vCU) is used to measure the resources consumed by read operations (such as search and query) and write operations (such as insert, upsert, and delete). The read and write costs vary for different vCU usage. The read and write costs vary for different vCU usage
Which type of CU should I pick?
Select the Performance-optimized CU if you instant search results and high concurrent traffic for real-time applications.
Choose the Capacity-optimized CU if you need to handle large vector datasets while maintaining reliable search speeds.
Opt for the Extended-capacity CU if you need to manage massive-scale datasets where optimizing total cost is prioritized over latency. Please contact sales if you need Extended-capacity CU.
How many CUs do I need for a given collection?
Performance-optimized CU: Supports up to 1.5 million 768-dimensional vectors.

Capacity-optimized CU: Supports up to 5 million 768-dimensional vectors.

Extended-capacity CU: Supports up to 20 million 768-dimensional vectors.

These estimates are based on vectors with primary keys only. Additional scalar fields like IDs or labels may reduce capacity. We recommend conducting your own tests for accurate assessment.
How can I get Zilliz Cloud discounts?
When you commit to an annual plan, you can get additional credits based on your usage.
How can I request a new cloud region?
To request a new cloud service provider region for Zilliz Cloud, please fill out the form.

Zilliz Cloud Pricing

Free

Serverless

Dedicated

BYOC