diff --git a/vllm/config.py b/vllm/config.py
index 7fbe04eaaf4f8..29f0839dcabba 100644
--- a/vllm/config.py
+++ b/vllm/config.py
@@ -513,11 +513,10 @@ class ModelConfig:
 
         # Reminder: Please update docs/source/usage/compatibility_matrix.rst
         # If the feature combo become valid
-        if device_config.device_type not in ("cuda", "tpu", "xpu", "hpu"):
+        if not current_platform.is_async_output_supported(self.enforce_eager):
             logger.warning(
-                "Async output processing is only supported for CUDA, TPU, XPU "
-                "and HPU."
-                "Disabling it for other platforms.")
+                "Async output processing is not supported on the "
+                "current platform type %s.", current_platform.device_type)
             self.use_async_output_proc = False
             return
 
@@ -527,16 +526,6 @@ class ModelConfig:
             self.use_async_output_proc = False
             return
 
-        # Reminder: Please update docs/source/usage/compatibility_matrix.rst
-        # If the feature combo become valid
-        if device_config.device_type == "cuda" and self.enforce_eager:
-            logger.warning(
-                "To see benefits of async output processing, enable CUDA "
-                "graph. Since, enforce-eager is enabled, async output "
-                "processor cannot be used")
-            self.use_async_output_proc = not self.enforce_eager
-            return
-
         # Async postprocessor is not necessary with embedding mode
         # since there is no token generation
         if self.task == "embedding":
diff --git a/vllm/platforms/cpu.py b/vllm/platforms/cpu.py
index 680ee74129739..e5142b985d1f2 100644
--- a/vllm/platforms/cpu.py
+++ b/vllm/platforms/cpu.py
@@ -1,4 +1,4 @@
-from typing import TYPE_CHECKING
+from typing import TYPE_CHECKING, Optional
 
 import psutil
 import torch
@@ -37,6 +37,10 @@ class CpuPlatform(Platform):
     def get_device_total_memory(cls, device_id: int = 0) -> int:
         return psutil.virtual_memory().total
 
+    @classmethod
+    def is_async_output_supported(cls, enforce_eager: Optional[bool]) -> bool:
+        return False
+
     @classmethod
     def inference_mode(cls):
         return torch.no_grad()
diff --git a/vllm/platforms/cuda.py b/vllm/platforms/cuda.py
index 846a1869da228..edaf377b501df 100644
--- a/vllm/platforms/cuda.py
+++ b/vllm/platforms/cuda.py
@@ -4,7 +4,7 @@ pynvml. However, it should not initialize cuda context.
 
 import os
 from functools import lru_cache, wraps
-from typing import TYPE_CHECKING, Callable, List, TypeVar
+from typing import TYPE_CHECKING, Callable, List, Optional, TypeVar
 
 import pynvml
 import torch
@@ -88,6 +88,16 @@ class CudaPlatformBase(Platform):
     def get_device_total_memory(cls, device_id: int = 0) -> int:
         raise NotImplementedError
 
+    @classmethod
+    def is_async_output_supported(cls, enforce_eager: Optional[bool]) -> bool:
+        if enforce_eager:
+            logger.warning(
+                "To see benefits of async output processing, enable CUDA "
+                "graph. Since, enforce-eager is enabled, async output "
+                "processor cannot be used")
+            return False
+        return True
+
     @classmethod
     def is_full_nvlink(cls, device_ids: List[int]) -> bool:
         raise NotImplementedError
diff --git a/vllm/platforms/hpu.py b/vllm/platforms/hpu.py
index 10aaa6d54962c..7f22bee3eaa74 100644
--- a/vllm/platforms/hpu.py
+++ b/vllm/platforms/hpu.py
@@ -1,4 +1,4 @@
-from typing import TYPE_CHECKING
+from typing import TYPE_CHECKING, Optional
 
 import torch
 
@@ -20,6 +20,10 @@ class HpuPlatform(Platform):
     def get_default_attn_backend(cls, selected_backend: _Backend) -> _Backend:
         return _Backend.HPU_ATTN
 
+    @classmethod
+    def is_async_output_supported(cls, enforce_eager: Optional[bool]) -> bool:
+        return True
+
     @staticmethod
     def inference_mode():
         return torch.no_grad()
diff --git a/vllm/platforms/interface.py b/vllm/platforms/interface.py
index 0be7df7941b8b..db06d2c18e681 100644
--- a/vllm/platforms/interface.py
+++ b/vllm/platforms/interface.py
@@ -6,11 +6,15 @@ from typing import TYPE_CHECKING, NamedTuple, Optional, Tuple, Union
 import numpy as np
 import torch
 
+from vllm.logger import init_logger
+
 if TYPE_CHECKING:
     from vllm.config import VllmConfig
 else:
     VllmConfig = None
 
+logger = init_logger(__name__)
+
 
 class _Backend(enum.Enum):
     FLASH_ATTN = enum.auto()
@@ -147,6 +151,13 @@ class Platform:
         """Get the total memory of a device in bytes."""
         raise NotImplementedError
 
+    @classmethod
+    def is_async_output_supported(cls, enforce_eager: Optional[bool]) -> bool:
+        """
+        Check if the current platform supports async output.
+        """
+        raise NotImplementedError
+
     @classmethod
     def inference_mode(cls):
         """A device-specific wrapper of `torch.inference_mode`.
diff --git a/vllm/platforms/neuron.py b/vllm/platforms/neuron.py
index 87655ea198303..1e5c4bddfa24f 100644
--- a/vllm/platforms/neuron.py
+++ b/vllm/platforms/neuron.py
@@ -1,4 +1,4 @@
-from typing import TYPE_CHECKING
+from typing import TYPE_CHECKING, Optional
 
 from .interface import Platform, PlatformEnum
 
@@ -18,6 +18,10 @@ class NeuronPlatform(Platform):
     def get_device_name(cls, device_id: int = 0) -> str:
         return "neuron"
 
+    @classmethod
+    def is_async_output_supported(cls, enforce_eager: Optional[bool]) -> bool:
+        return False
+
     @classmethod
     def check_and_update_config(cls, vllm_config: VllmConfig) -> None:
         parallel_config = vllm_config.parallel_config
diff --git a/vllm/platforms/openvino.py b/vllm/platforms/openvino.py
index 29b61e955d9ab..e0f8e8b4b49fe 100644
--- a/vllm/platforms/openvino.py
+++ b/vllm/platforms/openvino.py
@@ -1,4 +1,4 @@
-from typing import TYPE_CHECKING
+from typing import TYPE_CHECKING, Optional
 
 import torch
 
@@ -37,6 +37,10 @@ class OpenVinoPlatform(Platform):
     def get_device_name(self, device_id: int = 0) -> str:
         return "openvino"
 
+    @classmethod
+    def is_async_output_supported(cls, enforce_eager: Optional[bool]) -> bool:
+        return False
+
     @classmethod
     def inference_mode(self):
         return torch.inference_mode(mode=True)
diff --git a/vllm/platforms/rocm.py b/vllm/platforms/rocm.py
index 3c14fbc179f69..66674e3ebe91f 100644
--- a/vllm/platforms/rocm.py
+++ b/vllm/platforms/rocm.py
@@ -1,6 +1,6 @@
 import os
 from functools import lru_cache
-from typing import TYPE_CHECKING
+from typing import TYPE_CHECKING, Optional
 
 import torch
 
@@ -72,6 +72,16 @@ class RocmPlatform(Platform):
         device_props = torch.cuda.get_device_properties(device_id)
         return device_props.total_memory
 
+    @classmethod
+    def is_async_output_supported(cls, enforce_eager: Optional[bool]) -> bool:
+        if enforce_eager:
+            logger.warning(
+                "To see benefits of async output processing, enable CUDA "
+                "graph. Since, enforce-eager is enabled, async output "
+                "processor cannot be used")
+            return False
+        return True
+
     @classmethod
     def check_and_update_config(cls, vllm_config: VllmConfig) -> None:
         parallel_config = vllm_config.parallel_config
diff --git a/vllm/platforms/tpu.py b/vllm/platforms/tpu.py
index b138f7e1c54c5..10d874349f36b 100644
--- a/vllm/platforms/tpu.py
+++ b/vllm/platforms/tpu.py
@@ -1,4 +1,4 @@
-from typing import TYPE_CHECKING
+from typing import TYPE_CHECKING, Optional
 
 import torch
 
@@ -35,6 +35,10 @@ class TpuPlatform(Platform):
     def get_device_total_memory(cls, device_id: int = 0) -> int:
         raise NotImplementedError
 
+    @classmethod
+    def is_async_output_supported(cls, enforce_eager: Optional[bool]) -> bool:
+        return True
+
     @classmethod
     def inference_mode(cls):
         return torch.no_grad()
diff --git a/vllm/platforms/xpu.py b/vllm/platforms/xpu.py
index 9665786f4c499..11dbd04d55671 100644
--- a/vllm/platforms/xpu.py
+++ b/vllm/platforms/xpu.py
@@ -1,4 +1,4 @@
-from typing import TYPE_CHECKING
+from typing import TYPE_CHECKING, Optional
 
 import torch
 
@@ -41,6 +41,10 @@ class XPUPlatform(Platform):
         device_props = torch.xpu.get_device_properties(device_id)
         return device_props.total_memory
 
+    @classmethod
+    def is_async_output_supported(cls, enforce_eager: Optional[bool]) -> bool:
+        return True
+
     @staticmethod
     def inference_mode():
         return torch.no_grad()