diff --git a/vllm_omni/diffusion/worker/diffusion_worker.py b/vllm_omni/diffusion/worker/diffusion_worker.py
index 62cddd4a6..db1a9fded 100644
--- a/vllm_omni/diffusion/worker/diffusion_worker.py
+++ b/vllm_omni/diffusion/worker/diffusion_worker.py
@@ -255,6 +255,7 @@ class DiffusionWorker:
             "Final IR op priority after setting vLLM-Omni overrides: %s", vllm_config.kernel_config.ir_op_priority
         )
         self.vllm_config = vllm_config
+        current_omni_platform.init_diffusion_worker_vllm_config(vllm_config)
 
         # Initialize distributed environment
         with (
diff --git a/vllm_omni/platforms/interface.py b/vllm_omni/platforms/interface.py
index 14aa93b20..f92db4997 100644
--- a/vllm_omni/platforms/interface.py
+++ b/vllm_omni/platforms/interface.py
@@ -136,6 +136,11 @@ class OmniPlatform(Platform):
         """
         return "vllm_omni.diffusion.worker.diffusion_model_runner.DiffusionModelRunner"
 
+    @classmethod
+    def init_diffusion_worker_vllm_config(cls, vllm_config: Any) -> None:
+        """Initialize platform-specific state for diffusion worker VllmConfig."""
+        return None
+
     @classmethod
     def get_torch_device(cls, local_rank: int | None = None) -> torch.device:
         raise NotImplementedError
diff --git a/vllm_omni/platforms/npu/platform.py b/vllm_omni/platforms/npu/platform.py
index 6f2eb99aa..e2a576cfd 100644
--- a/vllm_omni/platforms/npu/platform.py
+++ b/vllm_omni/platforms/npu/platform.py
@@ -57,6 +57,12 @@ class NPUOmniPlatform(OmniPlatform, NPUPlatform):
     def get_omni_generation_worker_cls(cls) -> str:
         return "vllm_omni.platforms.npu.worker.npu_generation_worker.NPUGenerationWorker"
 
+    @classmethod
+    def init_diffusion_worker_vllm_config(cls, vllm_config: Any) -> None:
+        from vllm_ascend.ascend_config import init_ascend_config
+
+        init_ascend_config(vllm_config)
+
     @classmethod
     def get_default_stage_config_path(cls) -> str:
         return "vllm_omni/platforms/npu/stage_configs"