Start for Free and Pay as You Go

Vector Search

Vector Search for Production-level Generative AI

Vector Search on Astra DB is the only vector database for building real-world, production-level AI applications on real-time data.

Supported Cloud Providers


Support Type


Support Goals
Low Response Time

24 hours

Normal Response Time

24 hours

High Response Time

4 hours

Urgent Response Time

1 hour


15% of Astra commitment amount

Please contact us for enhanced support needs, migrations or other help you might need.

Astra DB can also be purchased through the AWS, Google Cloud, or Microsoft Azure marketplaces.

Astra Starter Pack

Overwhelmed by the maze of tools for building apps? Astra Starter Pack gives you what you need to start building and get into production—fast.

Metering Details


Read requests

A “Read Request Unit” (RRU) is the unit of measure for billing database reads.

It is based on the payload of each read query response.

A read request that returns up to 4KB of data is considered one RRU. If the request returns more than 4KB of data, additional RRUs are required. If the read request involves an ALLOW FILTERING query the data is measured prior to in-memory filtering.

Write requests

A “Write Request Unit” (WRU) is the unit of measure for billing database writes.

It is based on the payload size of each write request.

A write request with up to 1KB of data or a DELETE request are each considered one WRU. If the request has more than 1KB of data, additional WRUs are required.

A write index SAI is considered as a write request. The write index SAI size (and the number of WRUs required) is based on the size of each indexed column (not the size of the index), regardless of the column type.

Note: Writes in a multi-region configuration are billed for each region written to.


“Data Storage” means all data stored into the Database including the data, indexes and metadata.

Storage is calculated based on logical storage consumption. Data Storage is measured in GB per month, using the maximum data stored in the given month for the usage calculation.

You are not billed extra for standard backups of your data (It is included in the base storage costs).

Data Transfer

“Data Transfer” means the transfer of Customer Data out of the database. Billable units and pricing vary depending on whether the Data Transfer occurs within the same region of a cloud provider network (“Data Transfer - Same Region”), across regions within the same cloud provider network (“Data Transfer - Cross Region Within Cloud Provider Network”), or leaves the cloud provider network over the internet (“Data Transfer - Internet”).

Data Transfer charges are measured in GB.




Private Endpoint

Creating a “Private Endpoint” allows users to securely connect applications and third party SaaS offerings to Astra DB.

Users are billed at a fixed rate per hour plus data transfer charges per GB. Data transfer charges using private endpoints are for the transfer of Customer Data into and out of the database.

Cloud providers will also bill for use of private endpoint capabilities.

Vector Dimensions

Vector dimensions are the number of elements in a vector.

Users are billed an incremental metering rate for reads and writes per million vector dimensions, over and above non-vector read and write rates. The vector dimension rate is specified in the metering unit rates table below.


“Multi-Region” means replicating the database between multiple cloud regions enabling users to write to and read from multiple regions.

Additional charges include network egress between regions which is billed per GB. This cost is based on the source region. Users will also incur a one-time data transfer charge to replicate all data to the newly added region.

Metering Unit Rates


Write Request Units (per 1M)




Read Request Units (per 1M)




Data Storage (GB/month)




Data Transfer (GB) - Same Region




Data Transfer (GB) - Cross Region Within Cloud Provider Network




Data Transfer (GB) - Internet




Private Endpoint ($0.01/hr and data transfer charge of $0.01/GB ingress and egress)

Vector Dimension Read & Writes (per 1M)




Please note that DataStax may charge a higher rate for a specific region (by cloud provider) than what is shown in the pricing table.

Based on $1/ purchased credit and DS metering schedule
See our full list of regions

Vector Pricing Estimator

Estimate your monthly costs for Vector on Astra DB

Max 8192
Max 100M
Max 100M

Grand Total* : $ 0.00 / Month

Price includes a $25 credit per month.

*Pricing estimator is an estimate only and subject to change, please reach out to sales for more information.

Vector Search Pricing Considerations

The basic principles remain the same when leveraging Vector Search in Astra DB. We charge for reads, writes, storage and data transfer but the way we calculate these charges changes slightly with Vector Search to take into account the number of dimensions of the vector. This is broken down below and highlights which of the Metering Details you can reference to get more details. Find Metering Details and their associated Rates here.

Reads are based on the size of the non-vector data that is processed in order to satisfy the request (see Read Requests), the indexes leveraged to find the vector and the number of dimensions of the vector (see Vector Dimensions).

Writes are based on the size of the data written (see Write Requests), the number of indexes required to be written (see Write Requests) and the number of dimensions of the vector (see Vector Dimensions).

Storage is the total (compressed) size of the data you’re storing which includes metadata, indexes and vectors (see Storage).

Data transfer is based on the number of availability zones and the composition of what is being written including metadata, indexes and vectors (see Data Transfer).

Multi-Region Pricing Considerations

Multi-regional databases follow the same principles of single regional databases in terms of charging for reads, writes, data transfer and storage. There are additional charges for multi-region databases which are outlined below and we also highlight which of the Metering Details you can reference to get more details. Find Metering Details and their associated Rates here.

Reads are only charged for the region in which the read was done.

Writes are performed to every region which has been deployed with associated charges (see Write Requests).

Data is replicated to each region deployed and storage is charged for each region (see Storage).

Data Transfer is charged for transfer of data between regions and the rate depends on the kind of data replication implemented (see Data Transfer).

Enterprise Consumption Models

Astra Serverless is the underlying architecture under all Astra DB solutions. This provides for a cloud native, microservices based architecture that scales to support a number of different consumption models.

On Demand

On Demand is the default Enterprise consumption model. Many customers value the ability to scale to zero and only pay for what you use which is supported by our On Demand consumption model. This allows you to manage your budget closely and have the value that Astra DB provides scale with your usage.

Provisioned Capacity

More mature enterprises also have other considerations and it’s worth bearing these in mind when choosing which consumption model best suits your needs when using Astra DB.

If you have workloads that have any of the following characteristics you may also want to consider the Provisioned Capacity consumption model. This is the preferred consumption model of more mature workloads that have either

Strict performance needs and your traffic profile meets any of the following criteria:

  • Intra-day volatility
  • Low throughput combined with low latency needs
  • Batch workloads

Or workloads with

  • An ‘always on’ base capacity requirement
  • Seasonal changes in capacity needs

For these kinds of workloads, we recommend talking to our team about the benefits of Provisioned Capacity.

Ready to see how Vector pricing will work for you?

Chat with an ExpertRequest a demo


Where do I see my invoice?

You receive your invoice via email once a month. You will be able to see your consumption in the Astra console.

What payment methods are accepted?

You can pay for Astra DB via credit card, through your preferred cloud marketplace, or purchase Astra DB service credits via a direct contract with DataStax.

What happens to my services if I run out of credits?

If you run out of credits, your database may be temporarily paused until you resolve your payment issue. If you are on an Enterprise plan, you will be notified and will have 30 days to resolve your payment issue or your database may be suspended.

What happens to my services if my credit card expires or is unable to transact?

You will be notified and will have 30 days to resolve your payment issue or your database will be suspended.

What types of vector data types are supported?

Astra DB supports the full breadth of vector data types / dimensionality as well as other structured and NoSQL data types.

Why is the On Demand pricing based on Vector dimensions?

Size of the vector dimension is a good indicator of the work required to complete the request. This provides transparency on the value we are providing vs the work done to complete the request.