diff --git a/extension/apple/ExecuTorch/Exported/ExecuTorchTensor.h b/extension/apple/ExecuTorch/Exported/ExecuTorchTensor.h
index 53193dc1f5c..52ae4277c36 100644
--- a/extension/apple/ExecuTorch/Exported/ExecuTorchTensor.h
+++ b/extension/apple/ExecuTorch/Exported/ExecuTorchTensor.h
@@ -980,4 +980,87 @@ __attribute__((deprecated("This API is experimental.")))
 
 @end
 
+#pragma mark - RandomNormal Category
+
+@interface ExecuTorchTensor (RandomNormal)
+
+/**
+ * Creates a tensor with random values drawn from a normal distribution,
+ * with full specification of shape, strides, data type, and shape dynamism.
+ *
+ * @param shape An NSArray of NSNumber objects representing the desired shape.
+ * @param strides An NSArray of NSNumber objects representing the desired strides.
+ * @param dataType An ExecuTorchDataType value specifying the element type.
+ * @param shapeDynamism An ExecuTorchShapeDynamism value specifying whether the shape is static or dynamic.
+ * @return A new ExecuTorchTensor instance filled with values from a normal distribution.
+ */
++ (instancetype)randomNormalTensorWithShape:(NSArray<NSNumber *> *)shape
+                                    strides:(NSArray<NSNumber *> *)strides
+                                   dataType:(ExecuTorchDataType)dataType
+                              shapeDynamism:(ExecuTorchShapeDynamism)shapeDynamism
+    NS_SWIFT_NAME(randn(shape:strides:dataType:shapeDynamism:));
+
+/**
+ * Creates a tensor with random values drawn from a normal distribution,
+ * with the specified shape and data type.
+ *
+ * @param shape An NSArray of NSNumber objects representing the desired shape.
+ * @param dataType An ExecuTorchDataType value specifying the element type.
+ * @param shapeDynamism An ExecuTorchShapeDynamism value specifying whether the shape is static or dynamic.
+ * @return A new ExecuTorchTensor instance filled with values from a normal distribution.
+ */
++ (instancetype)randomNormalTensorWithShape:(NSArray<NSNumber *> *)shape
+                                   dataType:(ExecuTorchDataType)dataType
+                              shapeDynamism:(ExecuTorchShapeDynamism)shapeDynamism
+    NS_SWIFT_NAME(randn(shape:dataType:shapeDynamism:));
+
+/**
+ * Creates a tensor with random values drawn from a normal distribution,
+ * with the specified shape (using dynamic bound shape) and data type.
+ *
+ * @param shape An NSArray of NSNumber objects representing the desired shape.
+ * @param dataType An ExecuTorchDataType value specifying the element type.
+ * @return A new ExecuTorchTensor instance filled with values from a normal distribution.
+ */
++ (instancetype)randomNormalTensorWithShape:(NSArray<NSNumber *> *)shape
+                                   dataType:(ExecuTorchDataType)dataType
+    NS_SWIFT_NAME(randn(shape:dataType:));
+
+/**
+ * Creates a tensor with random normal values similar to an existing tensor,
+ * with the specified data type and shape dynamism.
+ *
+ * @param tensor An existing ExecuTorchTensor instance whose shape and strides are used.
+ * @param dataType An ExecuTorchDataType value specifying the desired element type.
+ * @param shapeDynamism An ExecuTorchShapeDynamism value specifying whether the shape is static or dynamic.
+ * @return A new ExecuTorchTensor instance filled with values from a normal distribution.
+ */
++ (instancetype)randomNormalTensorLikeTensor:(ExecuTorchTensor *)tensor
+                                    dataType:(ExecuTorchDataType)dataType
+                               shapeDynamism:(ExecuTorchShapeDynamism)shapeDynamism
+    NS_SWIFT_NAME(randn(like:dataType:shapeDynamism:));
+
+/**
+ * Creates a tensor with random normal values similar to an existing tensor,
+ * with the specified data type.
+ *
+ * @param tensor An existing ExecuTorchTensor instance whose shape and strides are used.
+ * @param dataType An ExecuTorchDataType value specifying the desired element type.
+ * @return A new ExecuTorchTensor instance filled with values from a normal distribution.
+ */
++ (instancetype)randomNormalTensorLikeTensor:(ExecuTorchTensor *)tensor
+                                    dataType:(ExecuTorchDataType)dataType
+    NS_SWIFT_NAME(randn(like:dataType:));
+
+/**
+ * Creates a tensor with random normal values similar to an existing tensor.
+ *
+ * @param tensor An existing ExecuTorchTensor instance.
+ * @return A new ExecuTorchTensor instance filled with values from a normal distribution.
+ */
++ (instancetype)randomNormalTensorLikeTensor:(ExecuTorchTensor *)tensor
+    NS_SWIFT_NAME(randn(like:));
+
+@end
+
 NS_ASSUME_NONNULL_END
diff --git a/extension/apple/ExecuTorch/Exported/ExecuTorchTensor.mm b/extension/apple/ExecuTorch/Exported/ExecuTorchTensor.mm
index 494618eb883..634a62c46f2 100644
--- a/extension/apple/ExecuTorch/Exported/ExecuTorchTensor.mm
+++ b/extension/apple/ExecuTorch/Exported/ExecuTorchTensor.mm
@@ -833,3 +833,61 @@ + (instancetype)randomTensorLikeTensor:(ExecuTorchTensor *)tensor {
 }
 
 @end
+
+@implementation ExecuTorchTensor (RandomNormal)
+
++ (instancetype)randomNormalTensorWithShape:(NSArray<NSNumber *> *)shape
+                                    strides:(NSArray<NSNumber *> *)strides
+                                   dataType:(ExecuTorchDataType)dataType
+                              shapeDynamism:(ExecuTorchShapeDynamism)shapeDynamism {
+  auto tensor = randn_strided(
+    utils::toVector<SizesType>(shape),
+    utils::toVector<StridesType>(strides),
+    static_cast<ScalarType>(dataType),
+    static_cast<TensorShapeDynamism>(shapeDynamism)
+  );
+  return [[self alloc] initWithNativeInstance:&tensor];
+}
+
++ (instancetype)randomNormalTensorWithShape:(NSArray<NSNumber *> *)shape
+                                   dataType:(ExecuTorchDataType)dataType
+                              shapeDynamism:(ExecuTorchShapeDynamism)shapeDynamism {
+  return [self randomNormalTensorWithShape:shape
+                                   strides:@[]
+                                  dataType:dataType
+                             shapeDynamism:shapeDynamism];
+}
+
++ (instancetype)randomNormalTensorWithShape:(NSArray<NSNumber *> *)shape
+                                   dataType:(ExecuTorchDataType)dataType {
+  return [self randomNormalTensorWithShape:shape
+                                   strides:@[]
+                                  dataType:dataType
+                             shapeDynamism:ExecuTorchShapeDynamismDynamicBound];
+}
+
++ (instancetype)randomNormalTensorLikeTensor:(ExecuTorchTensor *)tensor
+                                    dataType:(ExecuTorchDataType)dataType
+                               shapeDynamism:(ExecuTorchShapeDynamism)shapeDynamism {
+  return [self randomNormalTensorWithShape:tensor.shape
+                                   strides:tensor.strides
+                                  dataType:dataType
+                             shapeDynamism:shapeDynamism];
+}
+
++ (instancetype)randomNormalTensorLikeTensor:(ExecuTorchTensor *)tensor
+                                    dataType:(ExecuTorchDataType)dataType {
+  return [self randomNormalTensorWithShape:tensor.shape
+                                   strides:tensor.strides
+                                  dataType:dataType
+                             shapeDynamism:tensor.shapeDynamism];
+}
+
++ (instancetype)randomNormalTensorLikeTensor:(ExecuTorchTensor *)tensor {
+  return [self randomNormalTensorWithShape:tensor.shape
+                                   strides:tensor.strides
+                                  dataType:tensor.dataType
+                             shapeDynamism:tensor.shapeDynamism];
+}
+
+@end
diff --git a/extension/apple/ExecuTorch/__tests__/TensorTest.swift b/extension/apple/ExecuTorch/__tests__/TensorTest.swift
index 468ec85d261..27a6cac7b1e 100644
--- a/extension/apple/ExecuTorch/__tests__/TensorTest.swift
+++ b/extension/apple/ExecuTorch/__tests__/TensorTest.swift
@@ -637,4 +637,22 @@ class TensorTest: XCTestCase {
     XCTAssertEqual(tensor.shape, other.shape)
     XCTAssertEqual(tensor.count, other.count)
   }
+
+ func testRandomNormal() {
+    let tensor = Tensor.randn(shape: [4], dataType: .double)
+    XCTAssertEqual(tensor.shape, [4])
+    XCTAssertEqual(tensor.count, 4)
+    tensor.bytes { pointer, count, dataType in
+      XCTAssertEqual(dataType, .double)
+      let buffer = UnsafeBufferPointer(start: pointer.assumingMemoryBound(to: Double.self), count: count)
+      XCTAssertEqual(buffer.count, 4)
+    }
+  }
+
+  func testRandomNormalLike() {
+    let other = Tensor.zeros(shape: [4], dataType: .float)
+    let tensor = Tensor.randn(like: other)
+    XCTAssertEqual(tensor.shape, other.shape)
+    XCTAssertEqual(tensor.count, other.count)
+  }
 }