diff --git a/docs/getting_started/installation/npu/npu.inc.md b/docs/getting_started/installation/npu/npu.inc.md
index 02d757dfc..0b18f77d4 100644
--- a/docs/getting_started/installation/npu/npu.inc.md
+++ b/docs/getting_started/installation/npu/npu.inc.md
@@ -19,6 +19,7 @@ export DEVICE1=/dev/davinci1
 export IMAGE=quay.io/ascend/vllm-ascend:v0.11.0rc2
 docker run --rm \
     --name vllm-omni-npu \
+    --shm-size=1g \
     --device $DEVICE0 \
     --device $DEVICE1 \
     --device /dev/davinci_manager \
diff --git a/examples/offline_inference/qwen_image/gradio_demo.py b/examples/offline_inference/qwen_image/gradio_demo.py
index 2c4979e97..f7196a4fd 100644
--- a/examples/offline_inference/qwen_image/gradio_demo.py
+++ b/examples/offline_inference/qwen_image/gradio_demo.py
@@ -5,7 +5,7 @@ import gradio as gr
 import torch
 
 from vllm_omni.entrypoints.omni import Omni
-from vllm_omni.utils.platform_utils import detect_device_type
+from vllm_omni.utils.platform_utils import detect_device_type, is_npu
 
 ASPECT_RATIOS: dict[str, tuple[int, int]] = {
     "1:1": (1328, 1328),
@@ -59,7 +59,14 @@ def parse_args() -> argparse.Namespace:
 
 @lru_cache(maxsize=1)
 def get_omni(model_name: str) -> Omni:
-    return Omni(model=model_name)
+    # Enable VAE memory optimizations on NPU
+    vae_use_slicing = is_npu()
+    vae_use_tiling = is_npu()
+    return Omni(
+        model=model_name,
+        vae_use_slicing=vae_use_slicing,
+        vae_use_tiling=vae_use_tiling,
+    )
 
 
 def build_demo(args: argparse.Namespace) -> gr.Blocks:
diff --git a/examples/offline_inference/qwen_image/text_to_image.py b/examples/offline_inference/qwen_image/text_to_image.py
index 4524fcefe..6261d8067 100644
--- a/examples/offline_inference/qwen_image/text_to_image.py
+++ b/examples/offline_inference/qwen_image/text_to_image.py
@@ -7,7 +7,7 @@ from pathlib import Path
 import torch
 
 from vllm_omni.entrypoints.omni import Omni
-from vllm_omni.utils.platform_utils import detect_device_type
+from vllm_omni.utils.platform_utils import detect_device_type, is_npu
 
 
 def parse_args() -> argparse.Namespace:
@@ -49,7 +49,15 @@ def main():
     device = detect_device_type()
     generator = torch.Generator(device=device).manual_seed(args.seed)
 
-    omni = Omni(model=args.model)
+    # Enable VAE memory optimizations on NPU
+    vae_use_slicing = is_npu()
+    vae_use_tiling = is_npu()
+
+    omni = Omni(
+        model=args.model,
+        vae_use_slicing=vae_use_slicing,
+        vae_use_tiling=vae_use_tiling,
+    )
     images = omni.generate(
         args.prompt,
         height=args.height,
diff --git a/vllm_omni/diffusion/data.py b/vllm_omni/diffusion/data.py
index 1b612264b..4c471a5b3 100644
--- a/vllm_omni/diffusion/data.py
+++ b/vllm_omni/diffusion/data.py
@@ -80,6 +80,10 @@ class OmniDiffusionConfig:
     vae_cpu_offload: bool = True
     pin_cpu_memory: bool = True
 
+    # VAE memory optimization parameters
+    vae_use_slicing: bool = False
+    vae_use_tiling: bool = False
+
     # STA (Sliding Tile Attention) parameters
     mask_strategy_file_path: str | None = None
     # STA_mode: STA_Mode = STA_Mode.STA_INFERENCE
diff --git a/vllm_omni/diffusion/registry.py b/vllm_omni/diffusion/registry.py
index 0d834b2a0..adf97b442 100644
--- a/vllm_omni/diffusion/registry.py
+++ b/vllm_omni/diffusion/registry.py
@@ -24,6 +24,13 @@ def initialize_model(
         module = importlib.import_module(module_name)
         model_class = getattr(module, cls_name)
         model = model_class(od_config=od_config, prefix=mod_relname)
+
+        # Configure VAE memory optimization settings from config
+        if hasattr(model.vae, "use_slicing"):
+            model.vae.use_slicing = od_config.vae_use_slicing
+        if hasattr(model.vae, "use_tiling"):
+            model.vae.use_tiling = od_config.vae_use_tiling
+
         return model
     else:
         raise ValueError(f"Model class {od_config.model_class_name} not found in diffusion model registry.")
diff --git a/vllm_omni/diffusion/worker/npu/npu_worker.py b/vllm_omni/diffusion/worker/npu/npu_worker.py
index d35d187b4..83a9fa892 100644
--- a/vllm_omni/diffusion/worker/npu/npu_worker.py
+++ b/vllm_omni/diffusion/worker/npu/npu_worker.py
@@ -5,7 +5,8 @@ import os
 
 import torch
 import zmq
-from vllm.config import VllmConfig, set_current_vllm_config
+from transformers import PretrainedConfig
+from vllm.config import ModelConfig, VllmConfig, set_current_vllm_config
 from vllm.distributed.device_communicators.shm_broadcast import MessageQueue
 from vllm.distributed.parallel_state import (
     init_distributed_environment,
@@ -53,7 +54,12 @@ class NPUWorker:
         torch.npu.set_device(device)
 
         # hack
-        vllm_config = VllmConfig()
+        # set hf_config to a fake one to avolid get attr error
+        class _FakePretrainedConfig(PretrainedConfig):
+            def __getattr__(self, name):
+                return "fake"
+
+        vllm_config = VllmConfig(model_config=ModelConfig(hf_config=_FakePretrainedConfig()))
         vllm_config.parallel_config.tensor_parallel_size = self.od_config.num_gpus
         set_current_vllm_config(vllm_config)