diff --git a/vllm_fl/__init__.py b/vllm_fl/__init__.py
index 4c5d40f..7dbe0c8 100644
--- a/vllm_fl/__init__.py
+++ b/vllm_fl/__init__.py
@@ -22,10 +22,17 @@ def register():
 def register_model():
     """Register the FL model."""
     from vllm import ModelRegistry
+    import vllm.model_executor.models.qwen3_next as qwen3_next_module
 
     try:
         from vllm_fl.models.qwen3_next import Qwen3NextForCausalLM  # noqa: F401
 
+        qwen3_next_module.Qwen3NextForCausalLM = Qwen3NextForCausalLM
+        logger.warning(
+            "Qwen3NextForCausalLM has been patched to use vllm_fl.models.qwen3_next, "
+            "original vLLM implementation is overridden"
+        )
+
         ModelRegistry.register_model(
             "Qwen3NextForCausalLM", "vllm_fl.models.qwen3_next:Qwen3NextForCausalLM"
         )
diff --git a/vllm_fl/worker/worker.py b/vllm_fl/worker/worker.py
index 5975f70..2db1926 100644
--- a/vllm_fl/worker/worker.py
+++ b/vllm_fl/worker/worker.py
@@ -21,6 +21,7 @@
 from vllm.distributed import (
     ensure_model_parallel_initialized,
     init_distributed_environment,
+    set_custom_all_reduce,
 )
 from vllm.distributed.ec_transfer import ensure_ec_transfer_initialized
 from vllm.distributed.kv_transfer import (
@@ -1055,6 +1056,7 @@ def init_worker_distributed_environment(
     """Initialize the distributed environment."""
     attention_config = vllm_config.attention_config
     parallel_config = vllm_config.parallel_config
+    set_custom_all_reduce(not parallel_config.disable_custom_all_reduce)
     from vllm.model_executor.layers.batch_invariant import init_batch_invariance
 
     init_batch_invariance(attention_config.backend)