Pricing overview
With the Amazon Kendra Index, you pay for storage units, query units and connectors. Once you provision a Kendra Index, you are charged for storage unit and query unit hours from the time an index is created until it is deleted.
Amazon Kendra offers three index types: GenAI Enterprise Edition, Basic Enterprise Edition and Basic Developer Edition. The GenAI Enterprise Edition delivers the highest accuracy by leveraging the latest information retrieval technologies and semantic models. It offers high-availability and is built for production workloads. For the best experience, we recommend using the GenAI Index. The Basic Enterprise Edition provides semantic search capabilities and offers a high-availability service suitable for production workloads. The Basic Developer Edition also offers semantic search capabilities, designed for building proof-of-concept solutions, but it is not recommended for production workloads.
Pricing table
All Kendra indexes starts with a base index capacity which includes some storage and query capacity. You can add more storage units or query units to the base index (except for the Basic Developer Edition) based on your needs. Note that once an index is created you will incur charges for it regardless of whether you have utilized any storage or query capacity within that index. You can delete the index to stop the charges. The below table shows the price for all the components.
| Components | GenAI Enterprise Edition | Basic Enterprise Edition | Basic Developer Edition |
|---|---|---|---|
| Base Index (1 Storage Unit + 1 Query unit) |
$0.32 per hour Includes - - Up to 20,000 documents or 200MB of extracted text, whichever comes first - 0.1 QPS (approx 8000 queries per day) |
$1.4 per hour Includes - - Up to 100,000 documents or 30GB of extracted text, whichever comes first - 0.1 QPS (approx 8000 queries per day) |
$1.125 per hour Includes - Up to 10,000 documents or 3GB of extracted text, whichever comes first - 0.05 QPS (approx 4000 queroes per day) |
| Storage Units | $0.25 per hour Includes up to 20,000 documents or 200MB of extracted text whichever comes first |
$0.7 per hour Includes up to 100,000 documents or 30GB of extracted text whichever comes first |
Not Applicable |
| Query Units | $0.07 per hour Includes 0.1 QPS (approx 8000 queries per day) |
$0.7 per hour Includes 0.1 QPS (approx 8000 queries per day) |
Not Applicable |
| Connectors | Flat price of $30 per index per month | - $0.35 per hours when syncing - $1 per 1M documents scanned while syncing |
- $0.35 per hours when syncing - $1 per 1M documents scanned while syncing |
- Extracted text refers to the size of the content after text is extracted from the document. For reference, 1 page with 500 words is roughly 2 KB of extracted text. The size of extracted is unrelated to the file size of the document. For example, a powerpoint with a size of 10MB may contain less than 5KB of extracted text.
- The query units are shared across query and retrieve APIs. Customers can use the retrieve API for their retrieval-augmented-generative AI applications and query API for enterprise search.
- Basic Developer Edition index does not support adding more storage or query units
- For the GenAI Enterprise Edition Index, the connector is charged at a flat rate of $30 per index per month which includes up to 500 hours of sync usage per month. For higher usage, you can request a limit increase though service quotas. The monthly charge applies to every calendar month in which you have a connector configured for an index, regardless of the number of hours it was active during that month. The charges are not prorated based on usage duration within the month.
Free trial
You can get started for free with the Basic Developer Edition or the GenAI Enterprise Edition, that provides free usage of up to 750 hours for the first 30 days. Connector usage does not qualify for free usage. If you exceed the free tier usage limits, you will be charged the rates shown in the table above for the resources you use.
Pricing examples
Example 1 with GenAI Index (no additional storage or query capacity)
You’re creating an index of 15,000 drug research documents from a content management system. You expect to have 7,000 searches per day. You are only using the Kendra ingestion API to sync your documents, which means you will not incur costs for connectors. You are in us-east-1. Your monthly cost, assuming you are using the GenAI Enterprise Edition Index, will be calculated as:
| Cost Type | Pricing | Usage and Cost |
|---|---|---|
| GenAI Enterprise Edition | $0.32 per hour | Up to 8k queries per day. Up to 100k documents $0.32 per hour x 720 hours/month = $230.4 |
| Total monthly cost = $230.4 |
Example 2 with GenAI Index (additional storage, query capacity and connector usage)
Let’s say you want to build an internal search and Q&A experience for your employees for data spread across sources like Atlassian Confluence cloud, Microsoft SharePoint online, Salesforce and Google Drive. Let’s say you have 200,000 documents across your data sources, and your user base would generate an estimated 25,000 searches per day (~0.28 QPS). When you create your Kendra GenAI Index you will need to provision a base index, 9 units of storage capacity which can support up to 200,000 documents and 2 additional unit of query capacity which will support up to 0.3 QPS capacity. Your monthly cost will be calculated as:
| Cost Type | Pricing | Usage and Cost |
|---|---|---|
| Base GenAI Index | $0.32 per hour | 0.32*24*30 = $230.4 per month |
Storage Unit |
$0.25 per hour per unit |
9 units 9*0.25*24*30 = $1620 per month |
Query Unit |
$0.07 per hour per unit |
2 units 1*0.07*24*30 = $100 per month |
Connectors |
$30 per index per month |
$30 per month |
Total Monthly Cost |
$1980 per month |
|
Example 3 with Basic Enterprise Edition Index (adding storage and query capacity):
Let’s assume you’d like to search 150,000 documents in your marketing and HR repository, and your user base would generate an estimated 12,000 searches per day. When you create your Kendra index you’ll need to index 150,000 documents, which will take approximately 2 hours* of running the connectors. You’ll need to keep the index up to date by running a daily connector sync to pick up any new or updated documents from that repository. Assume this daily sync picks up 1,000 new or updated documents and runs for 30 minutes each day. That means the connector will be running for a total of 15 hours per month. Your monthly cost will be calculated as:
| Cost Type | Pricing | Usage |
|---|---|---|
| Base Basic Enterprise Edition | $1.4 per hour | Up to 8k queries per day. Up to 100k documents $1.4 per hour x 720 hours/month = $1,008 |
| Storage capacity | $0.7 per hour | +100k documents $0.7 per hour x 720 hours/month = $504 |
| Query capacity | $0.7 per hour | + up to 8k queries per day $0.7 per hour x 720 hours/month = $504 |
| Connector(s) | $0.000001 per scanned document $0.35 per hour while connectors are running |
First sync: 1M scanned docs x $0.000001 = $1 Incremental syncs: 1,000 scanned docs x 28 days x $0.000001 = $0.028 Sync time: 62 hours x $0.35 per hour = $21.70 Total cost for connectors = $22.728 |
| Total monthly cost | $2,038.78 | |
* Sync time will vary based on data source performance and average document size.