# CfnInferenceExperiment Creates an inference experiment using the configurations specified in the request. Use this API to setup and schedule an experiment to compare model variants on a Amazon SageMaker inference endpoint. For more information about inference experiments, see [Shadow tests](https://docs.aws.amazon.com/sagemaker/latest/dg/shadow-tests.html) . Amazon SageMaker begins your experiment at the scheduled time and routes traffic to your endpoint's model variants based on your specified configuration. While the experiment is in progress or after it has concluded, you can view metrics that compare your model variants. For more information, see [View, monitor, and edit shadow tests](https://docs.aws.amazon.com/sagemaker/latest/dg/shadow-tests-view-monitor-edit.html) . Extends: Construct › CfnElement › CfnRefElement › CfnResource › CfnInferenceExperiment Implements: IInspectable, ITaggable, IEnvironmentAware, IInferenceExperimentRef, IConstruct, IDependable - CloudFormation: [`AWS::SageMaker::InferenceExperiment`](https://docs.aws.amazon.com/AWSCloudFormation/latest/TemplateReference/aws-resource-sagemaker-inferenceexperiment.html) - Source: Generated code - Type last changed: v2.252.0 · 30 Apr 2026 ## Example ```ts // The code below shows an example of how to instantiate this type. // The values are placeholders you should change. import { aws_sagemaker as sagemaker } from 'aws-cdk-lib'; const cfnInferenceExperiment = new sagemaker.CfnInferenceExperiment(this, 'MyCfnInferenceExperiment', { endpointName: 'endpointName', modelVariants: [{ infrastructureConfig: { infrastructureType: 'infrastructureType', realTimeInferenceConfig: { instanceCount: 123, instanceType: 'instanceType', }, }, modelName: 'modelName', variantName: 'variantName', }], name: 'name', roleArn: 'roleArn', type: 'type', // the properties below are optional dataStorageConfig: { destination: 'destination', // the properties below are optional contentType: { csvContentTypes: ['csvContentTypes'], jsonContentTypes: ['jsonContentTypes'], }, kmsKey: 'kmsKey', }, description: 'description', desiredState: 'desiredState', kmsKey: 'kmsKey', schedule: { endTime: 'endTime', startTime: 'startTime', }, shadowModeConfig: { shadowModelVariants: [{ samplingPercentage: 123, shadowModelVariantName: 'shadowModelVariantName', }], sourceModelVariantName: 'sourceModelVariantName', }, statusReason: 'statusReason', tags: [{ key: 'key', value: 'value', }], }); ``` ## Initializer ```ts new CfnInferenceExperiment(scope: Construct, id: string, props: CfnInferenceExperimentProps) ``` ## Construct Props - `dataStorageConfig`: [IResolvable](/cdk/api/v2/docs/aws-cdk-lib.IResolvable.html) | [CfnInferenceExperiment.DataStorageConfigProperty](/cdk/api/v2/docs/aws-cdk-lib.aws_sagemaker.CfnInferenceExperiment.DataStorageConfigProperty.html) — The Amazon S3 location and configuration for storing inference request and response data. - `description`: string — The description of the inference experiment. - `desiredState`: string — The desired state of the experiment after stopping. The possible states are the following:. - `Completed` : The experiment completed successfully - `Cancelled` : The experiment was canceled - `endpointName`: string (required) — The name of the endpoint. - `kmsKey`: string — The AWS Key Management Service key that Amazon SageMaker uses to encrypt captured data at rest using Amazon S3 server-side encryption. - `modelVariants`: [IResolvable](/cdk/api/v2/docs/aws-cdk-lib.IResolvable.html) | [IResolvable](/cdk/api/v2/docs/aws-cdk-lib.IResolvable.html) | [CfnInferenceExperiment.ModelVariantConfigProperty](/cdk/api/v2/docs/aws-cdk-lib.aws_sagemaker.CfnInferenceExperiment.ModelVariantConfigProperty.html)[] (required) — An array of `ModelVariantConfigSummary` objects. There is one for each variant in the inference experiment. Each `ModelVariantConfigSummary` object in the array describes the infrastructure configuration for deploying the corresponding variant. - `name`: string (required) — The name of the inference experiment. - `roleArn`: string (required) — The ARN of the IAM role that Amazon SageMaker can assume to access model artifacts and container images, and manage Amazon SageMaker Inference endpoints for model deployment. - `schedule`: [IResolvable](/cdk/api/v2/docs/aws-cdk-lib.IResolvable.html) | [CfnInferenceExperiment.InferenceExperimentScheduleProperty](/cdk/api/v2/docs/aws-cdk-lib.aws_sagemaker.CfnInferenceExperiment.InferenceExperimentScheduleProperty.html) — The duration for which the inference experiment ran or will run. The maximum duration that you can set for an inference experiment is 30 days. - `shadowModeConfig`: [IResolvable](/cdk/api/v2/docs/aws-cdk-lib.IResolvable.html) | [CfnInferenceExperiment.ShadowModeConfigProperty](/cdk/api/v2/docs/aws-cdk-lib.aws_sagemaker.CfnInferenceExperiment.ShadowModeConfigProperty.html) — The configuration of `ShadowMode` inference experiment type, which shows the production variant that takes all the inference requests, and the shadow variant to which Amazon SageMaker replicates a percentage of the inference requests. For the shadow variant it also shows the percentage of requests that Amazon SageMaker replicates. - `statusReason`: string — The error message for the inference experiment status result. - `tags`: [CfnTag](/cdk/api/v2/docs/aws-cdk-lib.CfnTag.html)[] — An array of key-value pairs to apply to this resource. For more information, see [Tag](https://docs.aws.amazon.com/AWSCloudFormation/latest/UserGuide/aws-properties-resource-tags.html) . - `type`: string (required) — The type of the inference experiment. ## Properties - `attrArn`: string — The ARN for your inference experiment. - `attrCreationTime`: string — The timestamp at which the inference experiment was created. - `attrEndpointMetadata`: [IResolvable](/cdk/api/v2/docs/aws-cdk-lib.IResolvable.html) — The metadata of the endpoint on which the inference experiment ran. - `attrEndpointMetadataEndpointConfigName`: string - `attrEndpointMetadataEndpointName`: string - `attrEndpointMetadataEndpointStatus`: string - `attrLastModifiedTime`: string — The timestamp at which you last modified the inference experiment. - `attrStatus`: string — The status of the inference experiment. The following are the possible statuses for an inference experiment:. - `Creating` - Amazon SageMaker is creating your experiment. - `Created` - Amazon SageMaker has finished the creation of your experiment and will begin the experiment at the scheduled time. - `Updating` - When you make changes to your experiment, your experiment shows as updating. - `Starting` - Amazon SageMaker is beginning your experiment. - `Running` - Your experiment is in progress. - `Stopping` - Amazon SageMaker is stopping your experiment. - `Completed` - Your experiment has completed. - `Cancelled` - When you conclude your experiment early using the [StopInferenceExperiment](https://docs.aws.amazon.com/sagemaker/latest/APIReference/API_StopInferenceExperiment.html) API, or if any operation fails with an unexpected error, it shows as cancelled. - `CFN_RESOURCE_TYPE_NAME`: string (static) — The CloudFormation resource type name for this resource class. - `cfnOptions`: [ICfnResourceOptions](/cdk/api/v2/docs/aws-cdk-lib.ICfnResourceOptions.html) — Options for this resource, such as condition, update policy etc. - `cfnResourceType`: string — AWS resource type. - `creationStack`: string[] - `dataStorageConfig`: [IResolvable](/cdk/api/v2/docs/aws-cdk-lib.IResolvable.html) | [CfnInferenceExperiment.DataStorageConfigProperty](/cdk/api/v2/docs/aws-cdk-lib.aws_sagemaker.CfnInferenceExperiment.DataStorageConfigProperty.html) — The Amazon S3 location and configuration for storing inference request and response data. - `description`: string — The description of the inference experiment. - `desiredState`: string — The desired state of the experiment after stopping. The possible states are the following:. - `endpointName`: string — The name of the endpoint. - `env`: [ResourceEnvironment](/cdk/api/v2/docs/aws-cdk-lib.interfaces.ResourceEnvironment.html) - `inferenceExperimentRef`: [InferenceExperimentReference](/cdk/api/v2/docs/aws-cdk-lib.interfaces.aws_sagemaker.InferenceExperimentReference.html) — A reference to a InferenceExperiment resource. - `kmsKey`: string — The AWS Key Management Service key that Amazon SageMaker uses to encrypt captured data at rest using Amazon S3 server-side encryption. - `logicalId`: string — The logical ID for this CloudFormation stack element. The logical ID of the element is calculated from the path of the resource node in the construct tree. To override this value, use `overrideLogicalId(newLogicalId)`. - `modelVariants`: [IResolvable](/cdk/api/v2/docs/aws-cdk-lib.IResolvable.html) | [IResolvable](/cdk/api/v2/docs/aws-cdk-lib.IResolvable.html) | [CfnInferenceExperiment.ModelVariantConfigProperty](/cdk/api/v2/docs/aws-cdk-lib.aws_sagemaker.CfnInferenceExperiment.ModelVariantConfigProperty.html)[] — An array of `ModelVariantConfigSummary` objects. - `name`: string — The name of the inference experiment. - `node`: [Node](/cdk/api/v2/docs/constructs.Node.html) — The tree node. - `ref`: string — Return a string that will be resolved to a CloudFormation `{ Ref }` for this element. If, by any chance, the intrinsic reference of a resource is not a string, you could coerce it to an IResolvable through `Lazy.any({ produce: resource.ref })`. - `roleArn`: string — The ARN of the IAM role that Amazon SageMaker can assume to access model artifacts and container images, and manage Amazon SageMaker Inference endpoints for model deployment. - `schedule`: [IResolvable](/cdk/api/v2/docs/aws-cdk-lib.IResolvable.html) | [CfnInferenceExperiment.InferenceExperimentScheduleProperty](/cdk/api/v2/docs/aws-cdk-lib.aws_sagemaker.CfnInferenceExperiment.InferenceExperimentScheduleProperty.html) — The duration for which the inference experiment ran or will run. - `shadowModeConfig`: [IResolvable](/cdk/api/v2/docs/aws-cdk-lib.IResolvable.html) | [CfnInferenceExperiment.ShadowModeConfigProperty](/cdk/api/v2/docs/aws-cdk-lib.aws_sagemaker.CfnInferenceExperiment.ShadowModeConfigProperty.html) — The configuration of `ShadowMode` inference experiment type, which shows the production variant that takes all the inference requests, and the shadow variant to which Amazon SageMaker replicates a percentage of the inference requests. - `stack`: [Stack](/cdk/api/v2/docs/aws-cdk-lib.Stack.html) — The stack in which this element is defined. CfnElements must be defined within a stack scope (directly or indirectly). - `statusReason`: string — The error message for the inference experiment status result. - `tags`: [TagManager](/cdk/api/v2/docs/aws-cdk-lib.TagManager.html) — Tag Manager which manages the tags for this resource. - `tagsRaw`: [CfnTag](/cdk/api/v2/docs/aws-cdk-lib.CfnTag.html)[] — An array of key-value pairs to apply to this resource. - `type`: string — The type of the inference experiment. ## Methods - `addDeletionOverride(path: string): void` — Syntactic sugar for `addOverride(path, undefined)`. **Parameters** - `path` — The path of the value to delete. - `addDependency(target: CfnResource): void` — Use `addResourceDependency` instead. This method has been renamed to `addResourceDependency` to more clearly set it apart from `construct.node.addDependency`. See the documentation of that function for more details. - `addDependsOn(target: CfnResource): void` — Use `addResourceDependency` instead. This can be used for resources across stacks (or nested stack) boundaries and the dependency will automatically be transferred to the relevant scope. This method has been renamed to `addResourceDependency`, which makes it more clear that this method operates at a different level from the construct-level `construct.node.addDependency()` mechanism. - `addMetadata(key: string, value: any): void` — Add a value to the CloudFormation Resource Metadata. - `addOverride(path: string, value: any): void` — Adds an override to the synthesized CloudFormation resource. To add a property override, either use `addPropertyOverride` or prefix `path` with "Properties." (i.e. `Properties.TopicName`). If the override is nested, separate each nested level using a dot (.) in the path parameter. If there is an array as part of the nesting, specify the index in the path. To include a literal `.` in the property name, prefix with a `\`. In most programming languages you will need to write this as `"\\."` because the `\` itself will need to be escaped. For example, ```typescript cfnResource.addOverride('Properties.GlobalSecondaryIndexes.0.Projection.NonKeyAttributes', ['myattribute']); cfnResource.addOverride('Properties.GlobalSecondaryIndexes.1.ProjectionType', 'INCLUDE'); ``` would add the overrides ```json "Properties": { "GlobalSecondaryIndexes": [ { "Projection": { "NonKeyAttributes": [ "myattribute" ] ... } ... }, { "ProjectionType": "INCLUDE" ... }, ] ... } ``` The `value` argument to `addOverride` will not be processed or translated in any way. Pass raw JSON values in here with the correct capitalization for CloudFormation. If you pass CDK classes or structs, they will be rendered with lowercased key names, and CloudFormation will reject the template. **Parameters** - `path` — - The path of the property, you can use dot notation to override values in complex types. - `value` — - The value. - `addPropertyDeletionOverride(propertyPath: string): void` — Adds an override that deletes the value of a property from the resource definition. **Parameters** - `propertyPath` — The path to the property. - `addPropertyOverride(propertyPath: string, value: any): void` — Adds an override to a resource property. Syntactic sugar for `addOverride("Properties.<...>", value)`. **Parameters** - `propertyPath` — The path of the property. - `value` — The value. - `addResourceDependency(target: CfnResource, reason?: string): void` — Indicates that this resource depends on another resource and cannot be provisioned unless the other resource has been successfully provisioned. This can be used for resources across stacks (or nested stack) boundaries and the dependency will automatically be transferred to the relevant scope. This method only adds dependencies between L1 resources. If you are looking for a generic construct-to-construct dependency mechanism that works for all constructs including L2s, use `construct.node.addDependency` instead. - `applyCrossStackReferenceStrength(strength: ReferenceStrength): void` — Sets the cross-stack reference strength for this resource. When set, any cross-stack reference to this resource will use the specified strength instead of the global default from the consuming stack's context. **Parameters** - `strength` — - The reference strength to use for this resource. - `applyRemovalPolicy(policy?: RemovalPolicy, options?: RemovalPolicyOptions): void` — Sets the deletion policy of the resource based on the removal policy specified. The Removal Policy controls what happens to this resource when it stops being managed by CloudFormation, either because you've removed it from the CDK application or because you've made a change that requires the resource to be replaced. The resource can be deleted (`RemovalPolicy.DESTROY`), or left in your AWS account for data recovery and cleanup later (`RemovalPolicy.RETAIN`). In some cases, a snapshot can be taken of the resource prior to deletion (`RemovalPolicy.SNAPSHOT`). A list of resources that support this policy can be found in the following link: - `arnForInferenceExperiment(resource: IInferenceExperimentRef): string` (static) - `cfnPropertyName(cdkPropertyName: string): string` - `fromInferenceExperimentArn(scope: Construct, id: string, arn: string): IInferenceExperimentRef` (static) — Creates a new IInferenceExperimentRef from an ARN. - `fromInferenceExperimentName(scope: Construct, id: string, inferenceExperimentName: string): IInferenceExperimentRef` (static) — Creates a new IInferenceExperimentRef from a inferenceExperimentName. - `getAtt(attributeName: string, typeHint?: ResolutionTypeHint): Reference` — Returns a token for an runtime attribute of this resource. Ideally, use generated attribute accessors (e.g. `resource.arn`), but this can be used for future compatibility in case there is no generated attribute. **Parameters** - `attributeName` — The name of the attribute. - `getMetadata(key: string): any` — Retrieve a value value from the CloudFormation Resource Metadata. - `inspect(inspector: TreeInspector): void` — Examines the CloudFormation resource and discloses attributes. **Parameters** - `inspector` — tree inspector to collect and process attributes. - `isCfnElement(x: any): boolean` (static) — Returns `true` if a construct is a stack element (i.e. part of the synthesized cloudformation template). Uses duck-typing instead of `instanceof` to allow stack elements from different versions of this library to be included in the same stack. **Returns:** The construct as a stack element or undefined if it is not a stack element. - `isCfnInferenceExperiment(x: any): boolean` (static) — Checks whether the given object is a CfnInferenceExperiment. - `isCfnResource(x: any): boolean` (static) — Check whether the given object is a CfnResource. - `isConstruct(x: any): boolean` (static) — Checks if `x` is a construct. Use this method instead of `instanceof` to properly detect `Construct` instances, even when the construct library is symlinked. Explanation: in JavaScript, multiple copies of the `constructs` library on disk are seen as independent, completely different libraries. As a consequence, the class `Construct` in each copy of the `constructs` library is seen as a different class, and an instance of one class will not test as `instanceof` the other class. `npm install` will not create installations like this, but users may manually symlink construct libraries together or use a monorepo tool: in those cases, multiple copies of the `constructs` library can be accidentally installed, and `instanceof` will behave unpredictably. It is safest to avoid using `instanceof`, and using this type-testing method instead. **Parameters** - `x` — Any object. **Returns:** true if `x` is an object created from a class which extends `Construct`. - `obtainDependencies(): Stack | CfnResource[]` — Retrieves an array of resources and stacks this resource depends on. For resources depended on directly, returns the `CfnResource` object. For dependencies on other stacks, returns the `Stack` object. The order of the array is not guaranteed. - `overrideLogicalId(newLogicalId: string): void` — Overrides the auto-generated logical ID with a specific ID. **Parameters** - `newLogicalId` — The new logical ID to use for this stack element. - `removeDependency(target: CfnResource): void` — Use `removeResourceDependency` instead This can be used for resources across stacks (including nested stacks) and the dependency will automatically be removed from the relevant scope. - `removeResourceDependency(target: CfnResource): void` — Indicates that this resource no longer depends on another resource. This can be used for resources across stacks (including nested stacks) and the dependency will automatically be removed from the relevant scope. - `replaceDependency(target: CfnResource, newTarget: CfnResource): void` — Replaces one dependency with another. **Parameters** - `target` — The dependency to replace. - `newTarget` — The new dependency to add. - `toString(): string` — Returns a string representation of this construct. **Returns:** a string representation of this resource - `with(...mixins: IMixin[]): IConstruct` — Applies one or more mixins to this construct. Mixins are applied in order. The list of constructs is captured at the start of the call, so constructs added by a mixin will not be visited. Use multiple `with()` calls if subsequent mixins should apply to added constructs.