diff --git a/.changelog/38085.txt b/.changelog/38085.txt
new file mode 100644
index 000000000000..524ec90bf03c
--- /dev/null
+++ b/.changelog/38085.txt
@@ -0,0 +1,3 @@
+```release-note:enhancement
+resource/aws_sagemaker_endpoint_configuration: Add `production_variants.inference_ami_version` and `shadow_production_variants.inference_ami_version` arguments
+```
\ No newline at end of file
diff --git a/internal/service/sagemaker/endpoint_configuration.go b/internal/service/sagemaker/endpoint_configuration.go
index a040c56c9a07..5c226f55959a 100644
--- a/internal/service/sagemaker/endpoint_configuration.go
+++ b/internal/service/sagemaker/endpoint_configuration.go
@@ -295,6 +295,12 @@ func ResourceEndpointConfiguration() *schema.Resource {
 							Optional: true,
 							ForceNew: true,
 						},
+						"inference_ami_version": {
+							Type:         schema.TypeString,
+							Optional:     true,
+							ForceNew:     true,
+							ValidateFunc: validation.StringInSlice(sagemaker.ProductionVariantInferenceAmiVersion_Values(), false),
+						},
 						"initial_instance_count": {
 							Type:         schema.TypeInt,
 							Optional:     true,
@@ -433,6 +439,12 @@ func ResourceEndpointConfiguration() *schema.Resource {
 							Optional: true,
 							ForceNew: true,
 						},
+						"inference_ami_version": {
+							Type:         schema.TypeString,
+							Optional:     true,
+							ForceNew:     true,
+							ValidateFunc: validation.StringInSlice(sagemaker.ProductionVariantInferenceAmiVersion_Values(), false),
+						},
 						"initial_instance_count": {
 							Type:         schema.TypeInt,
 							Optional:     true,
@@ -697,6 +709,10 @@ func expandProductionVariants(configured []interface{}) []*sagemaker.ProductionV
 			l.EnableSSMAccess = aws.Bool(v)
 		}
 
+		if v, ok := data["inference_ami_version"].(string); ok && v != "" {
+			l.InferenceAmiVersion = aws.String(v)
+		}
+
 		containers = append(containers, l)
 	}
 
@@ -750,6 +766,10 @@ func flattenProductionVariants(list []*sagemaker.ProductionVariant) []map[string
 			l["enable_ssm_access"] = aws.BoolValue(i.EnableSSMAccess)
 		}
 
+		if i.InferenceAmiVersion != nil {
+			l["inference_ami_version"] = aws.StringValue(i.InferenceAmiVersion)
+		}
+
 		result = append(result, l)
 	}
 	return result
diff --git a/internal/service/sagemaker/endpoint_configuration_test.go b/internal/service/sagemaker/endpoint_configuration_test.go
index 65891f987f6f..9ce6f8e9c33b 100644
--- a/internal/service/sagemaker/endpoint_configuration_test.go
+++ b/internal/service/sagemaker/endpoint_configuration_test.go
@@ -214,6 +214,34 @@ func TestAccSageMakerEndpointConfiguration_ProductionVariants_serverless(t *test
 	})
 }
 
+func TestAccSageMakerEndpointConfiguration_ProductionVariants_ami(t *testing.T) {
+	ctx := acctest.Context(t)
+	rName := sdkacctest.RandomWithPrefix(acctest.ResourcePrefix)
+	resourceName := "aws_sagemaker_endpoint_configuration.test"
+
+	resource.ParallelTest(t, resource.TestCase{
+		PreCheck:                 func() { acctest.PreCheck(ctx, t) },
+		ErrorCheck:               acctest.ErrorCheck(t, names.SageMakerServiceID),
+		ProtoV5ProviderFactories: acctest.ProtoV5ProviderFactories,
+		CheckDestroy:             testAccCheckEndpointConfigurationDestroy(ctx),
+		Steps: []resource.TestStep{
+			{
+				Config: testAccEndpointConfigurationConfig_ami(rName),
+				Check: resource.ComposeTestCheckFunc(
+					testAccCheckEndpointConfigurationExists(ctx, resourceName),
+					resource.TestCheckResourceAttr(resourceName, "production_variants.#", acctest.Ct1),
+					resource.TestCheckResourceAttr(resourceName, "production_variants.0.inference_ami_version", "al2-ami-sagemaker-inference-gpu-2"), //lintignore:AWSAT002
+				),
+			},
+			{
+				ResourceName:      resourceName,
+				ImportState:       true,
+				ImportStateVerify: true,
+			},
+		},
+	})
+}
+
 func TestAccSageMakerEndpointConfiguration_ProductionVariants_serverlessProvisionedConcurrency(t *testing.T) {
 	ctx := acctest.Context(t)
 	rName := sdkacctest.RandomWithPrefix(acctest.ResourcePrefix)
@@ -1264,6 +1292,24 @@ resource "aws_sagemaker_endpoint_configuration" "test" {
 `, rName))
 }
 
+func testAccEndpointConfigurationConfig_ami(rName string) string {
+	//lintignore:AWSAT002
+	return acctest.ConfigCompose(testAccEndpointConfigurationConfig_base(rName), fmt.Sprintf(`
+resource "aws_sagemaker_endpoint_configuration" "test" {
+  name = %[1]q
+
+  production_variants {
+    variant_name           = "variant-1"
+    model_name             = aws_sagemaker_model.test.name
+    inference_ami_version  = "al2-ami-sagemaker-inference-gpu-2"
+    instance_type          = "ml.t2.medium"
+    initial_instance_count = 2
+    initial_variant_weight = 1
+  }
+}
+`, rName))
+}
+
 func testAccEndpointConfigurationConfig_serverlessProvisionedConcurrency(rName string) string {
 	return acctest.ConfigCompose(testAccEndpointConfigurationConfig_base(rName), fmt.Sprintf(`
 resource "aws_sagemaker_endpoint_configuration" "test" {
diff --git a/website/docs/r/sagemaker_endpoint_configuration.html.markdown b/website/docs/r/sagemaker_endpoint_configuration.html.markdown
index 75e799d31a97..8fa57a4c32f2 100644
--- a/website/docs/r/sagemaker_endpoint_configuration.html.markdown
+++ b/website/docs/r/sagemaker_endpoint_configuration.html.markdown
@@ -50,6 +50,7 @@ This resource supports the following arguments:
 * `container_startup_health_check_timeout_in_seconds` - (Optional) The timeout value, in seconds, for your inference container to pass health check by SageMaker Hosting. For more information about health check, see [How Your Container Should Respond to Health Check (Ping) Requests](https://docs.aws.amazon.com/sagemaker/latest/dg/your-algorithms-inference-code.html#your-algorithms-inference-algo-ping-requests). Valid values between `60` and `3600`.
 * `core_dump_config` - (Optional) Specifies configuration for a core dump from the model container when the process crashes. Fields are documented below.
 * `enable_ssm_access` - (Optional) You can use this parameter to turn on native Amazon Web Services Systems Manager (SSM) access for a production variant behind an endpoint. By default, SSM access is disabled for all production variants behind an endpoints.
+* `inference_ami_version` - (Optional) Specifies an option from a collection of preconfigured Amazon Machine Image (AMI) images. Each image is configured by Amazon Web Services with a set of software and driver versions. Amazon Web Services optimizes these configurations for different machine learning workloads.
 * `initial_instance_count` - (Optional) Initial number of instances used for auto-scaling.
 * `instance_type` - (Optional)  The type of instance to start.
 * `initial_variant_weight` - (Optional) Determines initial traffic distribution among all of the models that you specify in the endpoint configuration. If unspecified, it defaults to `1.0`.