diff --git a/vllm/_custom_ops.py b/vllm/_custom_ops.py
index e351d602189e2..2c09ca2c1407c 100644
--- a/vllm/_custom_ops.py
+++ b/vllm/_custom_ops.py
@@ -1,6 +1,6 @@
 import contextlib
 import functools
-from typing import List, Optional, Tuple, Type
+from typing import List, Optional, Tuple, Union
 
 import torch
 
@@ -336,7 +336,7 @@ def scaled_fp8_quant(
     """
     # This code assumes batch_dim and num_tokens are flattened
     assert (input.ndim == 2)
-    shape = input.shape
+    shape: Union[Tuple[int, int], torch.Size] = input.shape
     if num_token_padding:
         shape = (max(num_token_padding, input.shape[0]), shape[1])
     output = torch.empty(shape, device=input.device, dtype=torch.float8_e4m3fn)
diff --git a/vllm/multimodal/base.py b/vllm/multimodal/base.py
index f13885ef0dab0..aefb5f438c5ad 100644
--- a/vllm/multimodal/base.py
+++ b/vllm/multimodal/base.py
@@ -53,9 +53,7 @@ class MultiModalInputs(_MultiModalInputsBase):
     """
 
     @staticmethod
-    def _try_concat(
-        tensors: List[NestedTensors],
-    ) -> Union[GenericSequence[NestedTensors], NestedTensors]:
+    def _try_concat(tensors: List[NestedTensors]) -> BatchedTensors:
         """
         If each input tensor in the batch has the same shape, return a single
         batched tensor; otherwise, return a list of :class:`NestedTensors` with
@@ -105,7 +103,7 @@ def batch(inputs_list: List["MultiModalInputs"]) -> BatchedTensorInputs:
         return {
             k: MultiModalInputs._try_concat(item_list)
             for k, item_list in item_lists.items()
-        }  # type: ignore
+        }
 
     @staticmethod
     def as_kwargs(