InvokeEndpoint
After you deploy a model into production using Amazon SageMaker hosting services, your client applications use this API to get inferences from the model hosted at the specified endpoint.
For an overview of Amazon SageMaker, see How It Works.
Amazon SageMaker strips all POST headers except those supported by the API. Amazon SageMaker might add additional headers. You should not rely on the behavior of headers outside those enumerated in the request syntax.
Calls to InvokeEndpoint
are authenticated by using AWS
Signature Version 4. For information, see Authenticating
Requests (AWS Signature Version 4) in the Amazon S3 API Reference.
A customer's model containers must respond to requests within 60 seconds. The model itself can have a maximum processing time of 60 seconds before responding to invocations. If your model is going to take 50-60 seconds of processing time, the SDK socket timeout should be set to be 70 seconds.
Note
Endpoints are scoped to an individual account, and are not public. The URL does not contain the account ID, but Amazon SageMaker determines the account ID from the authentication token that is supplied by the caller.
Request Syntax
POST /endpoints/EndpointName
/invocations HTTP/1.1
Content-Type: ContentType
Accept: Accept
X-Amzn-SageMaker-Custom-Attributes: CustomAttributes
X-Amzn-SageMaker-Target-Model: TargetModel
X-Amzn-SageMaker-Target-Variant: TargetVariant
X-Amzn-SageMaker-Target-Container-Hostname: TargetContainerHostname
X-Amzn-SageMaker-Inference-Id: InferenceId
X-Amzn-SageMaker-Enable-Explanations: EnableExplanations
X-Amzn-SageMaker-Inference-Component: InferenceComponentName
X-Amzn-SageMaker-Session-Id: SessionId
Body
URI Request Parameters
The request uses the following URI parameters.
- Accept
-
The desired MIME type of the inference response from the model container.
Length Constraints: Maximum length of 1024.
Pattern:
\p{ASCII}*
- ContentType
-
The MIME type of the input data in the request body.
Length Constraints: Maximum length of 1024.
Pattern:
\p{ASCII}*
- CustomAttributes
-
Provides additional information about a request for an inference submitted to a model hosted at an Amazon SageMaker endpoint. The information is an opaque value that is forwarded verbatim. You could use this value, for example, to provide an ID that you can use to track a request or to provide other metadata that a service endpoint was programmed to process. The value must consist of no more than 1024 visible US-ASCII characters as specified in Section 3.3.6. Field Value Components
of the Hypertext Transfer Protocol (HTTP/1.1). The code in your model is responsible for setting or updating any custom attributes in the response. If your code does not set this value in the response, an empty value is returned. For example, if a custom attribute represents the trace ID, your model can prepend the custom attribute with
Trace ID:
in your post-processing function.This feature is currently supported in the AWS SDKs but not in the Amazon SageMaker Python SDK.
Length Constraints: Maximum length of 1024.
Pattern:
\p{ASCII}*
- EnableExplanations
-
An optional JMESPath expression used to override the
EnableExplanations
parameter of theClarifyExplainerConfig
API. See the EnableExplanations section in the developer guide for more information.Length Constraints: Minimum length of 1. Maximum length of 64.
Pattern:
.*
- EndpointName
-
The name of the endpoint that you specified when you created the endpoint using the CreateEndpoint API.
Length Constraints: Maximum length of 63.
Pattern:
^[a-zA-Z0-9](-*[a-zA-Z0-9])*
Required: Yes
- InferenceComponentName
-
If the endpoint hosts one or more inference components, this parameter specifies the name of inference component to invoke.
Length Constraints: Maximum length of 63.
Pattern:
^[a-zA-Z0-9]([\-a-zA-Z0-9]*[a-zA-Z0-9])?$
- InferenceId
-
If you provide a value, it is added to the captured data when you enable data capture on the endpoint. For information about data capture, see Capture Data.
Length Constraints: Minimum length of 1. Maximum length of 64.
Pattern:
\A\S[\p{Print}]*\z
- SessionId
-
Creates a stateful session or identifies an existing one. You can do one of the following:
-
Create a stateful session by specifying the value
NEW_SESSION
. -
Send your request to an existing stateful session by specifying the ID of that session.
With a stateful session, you can send multiple requests to a stateful model. When you create a session with a stateful model, the model must create the session ID and set the expiration time. The model must also provide that information in the response to your request. You can get the ID and timestamp from the
NewSessionId
response parameter. For any subsequent request where you specify that session ID, SageMaker routes the request to the same instance that supports the session.Length Constraints: Maximum length of 256.
Pattern:
^(NEW_SESSION)$|^[a-zA-Z0-9](-*[a-zA-Z0-9])*$
-
- TargetContainerHostname
-
If the endpoint hosts multiple containers and is configured to use direct invocation, this parameter specifies the host name of the container to invoke.
Length Constraints: Maximum length of 63.
Pattern:
^[a-zA-Z0-9](-*[a-zA-Z0-9])*
- TargetModel
-
The model to request for inference when invoking a multi-model endpoint.
Length Constraints: Minimum length of 1. Maximum length of 1024.
Pattern:
\A\S[\p{Print}]*\z
- TargetVariant
-
Specify the production variant to send the inference request to when invoking an endpoint that is running two or more variants. Note that this parameter overrides the default behavior for the endpoint, which is to distribute the invocation traffic based on the variant weights.
For information about how to use variant targeting to perform a/b testing, see Test models in production
Length Constraints: Maximum length of 63.
Pattern:
^[a-zA-Z0-9](-*[a-zA-Z0-9])*
Request Body
The request accepts the following binary data.
- Body
-
Provides input data, in the format specified in the
ContentType
request header. Amazon SageMaker passes all of the data in the body to the model.For information about the format of the request body, see Common Data Formats-Inference.
Length Constraints: Maximum length of 6291456.
Required: Yes
Response Syntax
HTTP/1.1 200
Content-Type: ContentType
x-Amzn-Invoked-Production-Variant: InvokedProductionVariant
X-Amzn-SageMaker-Custom-Attributes: CustomAttributes
X-Amzn-SageMaker-New-Session-Id: NewSessionId
X-Amzn-SageMaker-Closed-Session-Id: ClosedSessionId
Body
Response Elements
If the action is successful, the service sends back an HTTP 200 response.
The response returns the following HTTP headers.
- ClosedSessionId
-
If you closed a stateful session with your request, the ID of that session.
Length Constraints: Maximum length of 256.
Pattern:
^[a-zA-Z0-9](-*[a-zA-Z0-9])*$
- ContentType
-
The MIME type of the inference returned from the model container.
Length Constraints: Maximum length of 1024.
Pattern:
\p{ASCII}*
- CustomAttributes
-
Provides additional information in the response about the inference returned by a model hosted at an Amazon SageMaker endpoint. The information is an opaque value that is forwarded verbatim. You could use this value, for example, to return an ID received in the
CustomAttributes
header of a request or other metadata that a service endpoint was programmed to produce. The value must consist of no more than 1024 visible US-ASCII characters as specified in Section 3.3.6. Field Value Componentsof the Hypertext Transfer Protocol (HTTP/1.1). If the customer wants the custom attribute returned, the model must set the custom attribute to be included on the way back. The code in your model is responsible for setting or updating any custom attributes in the response. If your code does not set this value in the response, an empty value is returned. For example, if a custom attribute represents the trace ID, your model can prepend the custom attribute with
Trace ID:
in your post-processing function.This feature is currently supported in the AWS SDKs but not in the Amazon SageMaker Python SDK.
Length Constraints: Maximum length of 1024.
Pattern:
\p{ASCII}*
- InvokedProductionVariant
-
Identifies the production variant that was invoked.
Length Constraints: Maximum length of 1024.
Pattern:
\p{ASCII}*
- NewSessionId
-
If you created a stateful session with your request, the ID and expiration time that the model assigns to that session.
Length Constraints: Maximum length of 256.
Pattern:
^[a-zA-Z0-9](-*[a-zA-Z0-9])*;\sExpires=[0-9]{4}-[0-9]{2}-[0-9]{2}T[0-9]{2}:[0-9]{2}:[0-9]{2}Z$
The response returns the following as the HTTP body.
- Body
-
Includes the inference provided by the model.
For information about the format of the response body, see Common Data Formats-Inference.
If the explainer is activated, the body includes the explanations provided by the model. For more information, see the Response section under Invoke the Endpoint in the Developer Guide.
Length Constraints: Maximum length of 6291456.
Errors
For information about the errors that are common to all actions, see Common Errors.
- InternalDependencyException
-
Your request caused an exception with an internal dependency. Contact customer support.
HTTP Status Code: 530
- InternalFailure
-
An internal failure occurred.
HTTP Status Code: 500
- ModelError
-
Model (owned by the customer in the container) returned 4xx or 5xx error code.
HTTP Status Code: 424
- ModelNotReadyException
-
Either a serverless endpoint variant's resources are still being provisioned, or a multi-model endpoint is still downloading or loading the target model. Wait and try your request again.
HTTP Status Code: 429
- ServiceUnavailable
-
The service is unavailable. Try your call again.
HTTP Status Code: 503
- ValidationError
-
Inspect your request and try again.
HTTP Status Code: 400
Examples
Pass a trace ID in the CustomAttribute of a request and return it in the CustomAttribute of the response.
In this example a trace ID is passed to the service endpoint in the
CustomAttributes
header of the request and then retrieved and
returned in the CustomAttributes
header of the response.
Sample Request
import boto3
client = boto3.client('sagemaker-runtime')
custom_attributes = "c000b4f9-df62-4c85-a0bf-7c525f9104a4" # An example of a trace ID.
endpoint_name = "..." # Your endpoint name.
content_type = "..." # The MIME type of the input data in the request body.
accept = "..." # The desired MIME type of the inference in the response.
payload = "..." # Payload for inference.
response = client.invoke_endpoint(
EndpointName=endpoint_name,
CustomAttributes=custom_attributes,
ContentType=content_type,
Accept=accept,
Body=payload
)
print(response['CustomAttributes']) # If model receives and updates the custom_attributes header
# by adding "Trace id: " in front of custom_attributes in the request,
# custom_attributes in response becomes
# "Trace ID: c000b4f9-df62-4c85-a0bf-7c525f9104a4"
Sample Response
Trace ID: c000b4f9-df62-4c85-a0bf-7c525f9104a4
See Also
For more information about using this API in one of the language-specific AWS SDKs, see the following: