diff --git a/stubs/tensorflow/tensorflow/keras/metrics.pyi b/stubs/tensorflow/tensorflow/keras/metrics.pyi
index 29f35ee56..577170d41 100644
--- a/stubs/tensorflow/tensorflow/keras/metrics.pyi
+++ b/stubs/tensorflow/tensorflow/keras/metrics.pyi
@@ -1,9 +1,120 @@
-from tensorflow import Tensor
-from tensorflow._aliases import TensorCompatible
+from _typeshed import Incomplete
+from abc import ABCMeta, abstractmethod
+from collections.abc import Callable, Iterable, Sequence
+from typing import Any, Literal
+from typing_extensions import Self, TypeAlias, override
 
+import tensorflow as tf
+from tensorflow import Operation, Tensor
+from tensorflow._aliases import DTypeLike, KerasSerializable, TensorCompatible
+from tensorflow.keras.initializers import _Initializer
+
+_Output: TypeAlias = Tensor | dict[str, Tensor]
+
+class Metric(tf.keras.layers.Layer[tf.Tensor, tf.Tensor], metaclass=ABCMeta):
+    def __init__(self, name: str | None = None, dtype: DTypeLike | None = None) -> None: ...
+    def __new__(cls, *args: Any, **kwargs: Any) -> Self: ...
+    def merge_state(self, metrics: Iterable[Self]) -> list[Operation]: ...
+    def reset_state(self) -> None: ...
+    @abstractmethod
+    def update_state(
+        self, y_true: TensorCompatible, y_pred: TensorCompatible, sample_weight: TensorCompatible | None = None
+    ) -> Operation | None: ...
+    @abstractmethod
+    def result(self) -> _Output: ...
+    # Metric inherits from keras.Layer, but its add_weight method is incompatible with the one from "Layer".
+    @override
+    def add_weight(  # type: ignore
+        self,
+        name: str,
+        shape: Iterable[int | None] | None = (),
+        aggregation: tf.VariableAggregation = ...,
+        synchronization: tf.VariableSynchronization = ...,
+        initializer: _Initializer | None = None,
+        dtype: DTypeLike | None = None,
+    ) -> None: ...
+
+class AUC(Metric):
+    _from_logits: bool
+    _num_labels: int
+    num_labels: int | None
+    def __init__(
+        self,
+        num_thresholds: int = 200,
+        curve: Literal["ROC", "PR"] = "ROC",
+        summation_method: Literal["interpolation", "minoring", "majoring"] = "interpolation",
+        name: str | None = None,
+        dtype: DTypeLike | None = None,
+        thresholds: Sequence[float] | None = None,
+        multi_label: bool = False,
+        num_labels: int | None = None,
+        label_weights: TensorCompatible | None = None,
+        from_logits: bool = False,
+    ) -> None: ...
+    def update_state(
+        self, y_true: TensorCompatible, y_pred: TensorCompatible, sample_weight: TensorCompatible | None = None
+    ) -> Operation: ...
+    def result(self) -> tf.Tensor: ...
+
+class Precision(Metric):
+    def __init__(
+        self,
+        thresholds: float | Sequence[float] | None = None,
+        top_k: int | None = None,
+        class_id: int | None = None,
+        name: str | None = None,
+        dtype: DTypeLike | None = None,
+    ) -> None: ...
+    def update_state(
+        self, y_true: TensorCompatible, y_pred: TensorCompatible, sample_weight: TensorCompatible | None = None
+    ) -> Operation: ...
+    def result(self) -> tf.Tensor: ...
+
+class Recall(Metric):
+    def __init__(
+        self,
+        thresholds: float | Sequence[float] | None = None,
+        top_k: int | None = None,
+        class_id: int | None = None,
+        name: str | None = None,
+        dtype: DTypeLike | None = None,
+    ) -> None: ...
+    def update_state(
+        self, y_true: TensorCompatible, y_pred: TensorCompatible, sample_weight: TensorCompatible | None = None
+    ) -> Operation: ...
+    def result(self) -> tf.Tensor: ...
+
+class MeanMetricWrapper(Metric):
+    def __init__(
+        self, fn: Callable[[tf.Tensor, tf.Tensor], tf.Tensor], name: str | None = None, dtype: DTypeLike | None = None
+    ) -> None: ...
+    def update_state(
+        self, y_true: TensorCompatible, y_pred: TensorCompatible, sample_weight: TensorCompatible | None = None
+    ) -> Operation: ...
+    def result(self) -> tf.Tensor: ...
+
+class BinaryAccuracy(MeanMetricWrapper):
+    def __init__(self, name: str | None = "binary_accuracy", dtype: DTypeLike | None = None, threshold: float = 0.5) -> None: ...
+
+class Accuracy(MeanMetricWrapper):
+    def __init__(self, name: str | None = "accuracy", dtype: DTypeLike | None = None) -> None: ...
+
+class CategoricalAccuracy(MeanMetricWrapper):
+    def __init__(self, name: str | None = "categorical_accuracy", dtype: DTypeLike | None = None) -> None: ...
+
+class TopKCategoricalAccuracy(MeanMetricWrapper):
+    def __init__(self, k: int = 5, name: str | None = "top_k_categorical_accuracy", dtype: DTypeLike | None = None) -> None: ...
+
+class SparseTopKCategoricalAccuracy(MeanMetricWrapper):
+    def __init__(
+        self, k: int = 5, name: str | None = "sparse_top_k_categorical_accuracy", dtype: DTypeLike | None = None
+    ) -> None: ...
+
+def serialize(metric: KerasSerializable, use_legacy_format: bool = False) -> dict[str, Any]: ...
 def binary_crossentropy(
     y_true: TensorCompatible, y_pred: TensorCompatible, from_logits: bool = False, label_smoothing: float = 0.0, axis: int = -1
 ) -> Tensor: ...
 def categorical_crossentropy(
     y_true: TensorCompatible, y_pred: TensorCompatible, from_logits: bool = False, label_smoothing: float = 0.0, axis: int = -1
 ) -> Tensor: ...
+def __getattr__(name: str) -> Incomplete: ...