AIOZ Storage Vector Database Pricing: What You Pay For

AIOZ Storage vector database pricing has five charges: upload at 0.04 USD per GB, storage at 0.05 USD per GB per month, a flat 0.000002 USD per query, a tiered charge for data searched, and 0.01 USD per GB of data returned. All of it comes out of the same pre-paid balance as regular object storage. This guide explains each line, works the arithmetic for a realistic index, and marks the one place where the docs leave a gap, so you know which part of the example is an assumption.

TL;DR:

• Documented rates: upload 0.04 USD per GB, storage 0.05 USD per GB per month (charged hourly), 0.000002 USD per query, data searched from 0.0000028 down to 0.0000002 USD per GB depending on index size, data returned 0.01 USD per GB with the first 512 KB of each query free.
• An idle index costs only its storage. For 1,000,000 vectors of 1,536 dimensions, that is about 0.31 USD per month.
• The docs do not say how "data searched" is measured. The example below assumes each query is charged for the whole index, which is a ceiling.
• When the balance is too low, upload and query requests are rejected.

What Does AIOZ Storage Charge for a Vector Database?

AIOZ's billing page lists these prices for the vector database:

The cost docs describe the contrast with provisioned services. A provisioned vector database bills for capacity around the clock, so a quiet index costs the same as a busy one. A usage-based one bills for what you store and what you ask, so cost follows activity. AIOZ's model is the second kind.

How the Data Searched Charge Works

Data searched is priced in steps, and the price per GB falls as the index grows:

A query is charged at the average price across all vectors in the index. AIOZ's own example: an index with 1,000,000 vectors has 100,000 at 0.0000028 and 900,000 at 0.0000014, so the average is 0.00000154 USD per GB searched. The same method gives about 0.000001414 USD for 10,000,000 vectors (100,000 at the first rate and 9,900,000 at the second).

What the docs do not state is how many GB a single query counts as searched. They say the price depends on the size of your index, which suggests the whole index, but that is an inference. Amazon S3 Vectors, which AWS prices in a similar three-part way (request fee, data processed, data returned), defines its data processed as the vectors in the index multiplied by their average size. If AIOZ measures the same way, the example below is exact. If it measures less, your bill is lower.

How Big Is a Vector?

AIOZ Storage supports float32 vectors, and a float32 number takes 4 bytes, so a vector takes 4 bytes per dimension. A 1,536-dimension vector, the size AIOZ's own guides use for OpenAI's text-embedding-3-small, is 6,144 bytes. Metadata and the key are stored with each vector and add to that. The docs do not say whether they count toward the billed GB, so the example uses the raw vector bytes only. If you are choosing a dimension, What Is an Embedding Model? explains what the number means for quality and size.

A Worked Monthly Example

Take 1,000,000 vectors of 1,536 dimensions and use 1 GB = 1,000,000,000 bytes, since the docs do not define the unit. The raw vector data is 6.144 GB.

That is about 1.45 USD per month, plus the one-time upload. At 10,000,000 queries a month, the flat fees become 20.00 USD and the data searched ceiling becomes 94.62 USD, for a total near 114.92 USD with storage. Two things follow. In this example storage is the small part of the bill, and query volume drives it. And the data searched line is the one worth measuring on your own account, because it is the uncertain one: run a day of real traffic, compare the balance change with the flat fees, and you will see how AIOZ counts it.

Memory Calls Add a Flat Fee

If you use Super Intelligence Memory instead of managing vectors yourself, each Record or Recall call adds 0.00005 USD on top of the vector charges. A Record also pays the upload and storage for the saved statement, and a Recall pays the query charges. A million Recall calls add 50 USD in memory fees alone. AIOZ Storage Super Intelligence Memory: Record and Recall covers the feature, and Vector Database vs Agent Memory on AIOZ Storage: A Guide compares the two paths on cost.

Keep the Balance Funded

Vector usage is charged from the same pre-paid balance as object storage. The billing page says that when the balance is too low, upload and query requests are rejected, so a retrieval pipeline can fail suddenly if the balance runs out. The general rules for minimum balance and account locking are in How AIOZ Storage Pricing Works: Storage and Delivery Costs. For an application that retrieves context for a language model, Object Storage Vector Database: Where AIOZ Storage Fits shows where those query calls come from.

Frequently Asked Questions

How much does the AIOZ Storage vector database cost per month?
It depends on size and traffic. Storage is 0.05 USD per GB per month, and each query costs 0.000002 USD plus a data searched charge and any data returned above 512 KB. For 1,000,000 vectors of 1,536 dimensions and 100,000 queries, the example above comes to about 1.45 USD a month under a whole-index assumption.

What is the data searched charge?
A per-GB price that falls as the index grows: 0.0000028 USD for the first 100,000 vectors, 0.0000014 USD up to 10,000,000, and 0.0000002 USD above that. A query is charged at the average across the whole index.

Is the first part of each query result free?
Yes. The first 512 KB returned by each query is free, and data returned beyond that costs 0.01 USD per GB.

What happens when my balance runs out?
Upload and query requests are rejected until you add funds.

Do Record and Recall calls cost extra?
Yes. Each call adds a flat 0.00005 USD on top of the underlying vector charges.

Does an idle index cost anything?
Only its storage, charged hourly. The docs describe the model as usage-based, so a quiet index does not pay for idle capacity.

Related Reading

• Vector Database vs Agent Memory on AIOZ Storage: A Guide - choosing between the two features, with a cost comparison.
• AIOZ Storage Super Intelligence Memory: Record and Recall - the memory fee in context.
• How AIOZ Storage Pricing Works: Storage and Delivery Costs - object storage pricing and penalties.
• What Is an Embedding Model? - where dimension and vector size come from.

References

• AIOZ Storage Docs: Billing
• AIOZ Storage Docs: Vector Database Cost, Use Cases and Next Steps
• AIOZ Storage Docs: Vectors and Vector Databases
• AIOZ Storage Docs: Super Intelligence Memory
• AWS: Amazon S3 Pricing, S3 Vectors
• AWS: Working With S3 Vectors and Vector Buckets