AWS Kendra cost: an always-on index with a high baseline
Kendra bills per index-hour by edition (Developer and Enterprise), running around the clock, plus connector scans and extra query or document capacity. The always-on index fee is a significant baseline. Here is when Kendra's intelligent search pays off.
Quick answer
Kendra bills per index-hour by edition, running continuously: the Developer edition is roughly $810/month and Enterprise several times more, plus connector scan hours and charges for extra query and document capacity. The always-on index fee is a significant fixed baseline, so Kendra suits high-value enterprise search where its ML-powered relevance justifies the cost, not small or low-traffic search needs.
Amazon Kendra provides machine-learning-powered enterprise search. Unlike per-query search services, Kendra runs a provisioned index billed by the hour, so it has a meaningful fixed cost whether or not anyone searches. That baseline is the defining feature of its pricing.
An always-on index
| Edition | Approximate monthly baseline |
|---|---|
| Developer | ~$810 (index runs continuously) |
| Enterprise | Several times higher |
The Developer edition index costs roughly $810 a month just to run, and Enterprise several times that. On top of the index, connector scans (crawling data sources) bill per hour, and exceeding included query and document capacity adds charges. The index baseline dominates for most deployments.
Where the baseline is justified
Kendra's value is ML-driven relevance, natural-language queries, and connectors across enterprise data sources, capabilities that a keyword search engine lacks. For a large organization with substantial content and high-value search needs (support, knowledge management), that relevance can justify the baseline. For small content sets or low search volume, the always-on fee is hard to justify, and a simpler, cheaper search (OpenSearch, or a per-query option) fits better.
Controlling Kendra cost
Use the Developer edition for proofs of concept and smaller deployments and Enterprise only when its capacity and features are needed, schedule connector scans to the frequency your data actually changes rather than constantly, right-size query and document capacity, and genuinely evaluate whether Kendra's intelligent search is required versus a cheaper engine. The always-on baseline means Kendra is worth it only when the search quality delivers proportional value, the tier-to-need discipline seen in other premium services like Apigee.
FAQ
How is AWS Kendra priced?
Per index-hour by edition, running continuously: the Developer edition is roughly $810 per month and Enterprise several times more, plus connector scan hours for crawling data sources and charges for exceeding included query and document capacity. The always-on index fee is the significant fixed baseline.
Why is AWS Kendra expensive?
Because it runs a provisioned, machine-learning-powered index billed by the hour whether or not anyone searches, so it has a high fixed baseline (roughly $810 a month for Developer, more for Enterprise). Unlike per-query search, an idle Kendra index still costs its hourly fee.
When is Kendra worth the cost?
For large organizations with substantial content and high-value search needs (support, knowledge management) where its ML-driven relevance, natural-language queries, and enterprise connectors deliver value a keyword engine cannot. For small content sets or low search volume, the always-on baseline is hard to justify.
How do I reduce Kendra cost?
Use the Developer edition for proofs of concept and smaller deployments, reserve Enterprise for when its capacity and features are needed, schedule connector scans to match how often data actually changes, right-size query and document capacity, and evaluate whether Kendra's intelligent search is required versus a cheaper engine like OpenSearch.
Kendra or OpenSearch for cost?
OpenSearch is cheaper and more flexible for general search, but requires you to build relevance. Kendra costs more (a high always-on baseline) but provides ML-powered relevance and natural-language search out of the box. Choose Kendra when that intelligent search quality justifies the baseline; OpenSearch when a simpler engine suffices.
Does C3X estimate Kendra cost?
C3X prices the Kendra index resource and its edition from your Terraform, so the always-on baseline appears before deploy. Connector scans and extra capacity depend on usage, which you provide as assumptions.
What to do next
Estimate the infrastructure around your search before you deploy. C3X reads your Terraform and prices your resources against a live catalog. Start with the quickstart.
Pricing references
Related posts
- Why is my AWS bill so high: finding the real cost drivers
A high AWS bill almost always traces to a few dominant drivers, over-provisioned compute, data transfer, idle resources, and unmanaged growth, rather than a mystery. Here is how to find what is driving your bill and bring it down.
- GCP Cloud Composer cost: a managed Airflow baseline that is always on
Cloud Composer bills environment compute (per-hour, always on) plus the underlying GKE, database, and storage. The always-on environment fee is significant, so consolidating DAGs and right-sizing the environment matter. Here is the breakdown.
- AWS Comprehend cost: per unit of text analyzed
Comprehend bills per unit of text (100 characters, with a per-request minimum) for NLP tasks like sentiment, entities, and key phrases, with higher rates for custom models. Analyzing more text or in more ways than needed drives the bill.
- AWS Personalize cost: training, ingestion, and inference capacity
Personalize bills data ingestion, training hours to build recommendation models, and real-time inference by provisioned throughput (TPS-hours). Over-provisioned recommendation capacity and frequent retraining drive the bill. Here is the model.
Share this post
Try C3X on your own Terraform
Free and open source. No API key required. One command to install, one command to estimate.