Amazon Kendra will no longer be open to new customers starting on July 30, 2026. If you would like to use the service, please sign up prior to July 30. For capabilities similar to Amazon Kendra, explore Amazon Bedrock Knowledge Bases. [Learn more](https://docs.aws.amazon.com/kendra/latest/dg/kendra-availability-change.html).

# Retrieving responses from indexes in Amazon Kendra
<a name="index-searching"></a>

After you create an index, you can start searching your documents.

To search an Amazon Kendra index, you use either the [ Retrieve](https://docs.aws.amazon.com/kendra/latest/APIReference/API_Retrieve.html) API operation or the [Query](https://docs.aws.amazon.com/kendra/latest/APIReference/API_Query.html) API operation.

The Retrieve API operation is ideal for Retrieval Augmented Generation (RAG) use cases. For a given query, it returns a ranked list of semantically relevant passages of up to 200 token words. You can send these to a large language model (LLM) to generate an answer using RAG. For more information, see [Searching an index](https://docs.aws.amazon.com/kendra/latest/dg/searching.html).

The Query API operation is best for document search use cases. For a given query, it returns a list of ranked documents with 100 word excerpts that are relevant to the query. This is useful for traditional document search use cases where users are browsing through a list of ranked documents.

To see what features are supported by the Retrieve and Query API operations for each index type, see [Index types](https://docs.aws.amazon.com/kendra/latest/dg/hiw-index-types.html).