diff --git a/vllm/config.py b/vllm/config.py
index 26edd4567b9ac..2eb5bdd18d812 100644
--- a/vllm/config.py
+++ b/vllm/config.py
@@ -531,6 +531,7 @@ class ParallelConfig:
             If None, will use synchronous tokenization.
         ray_workers_use_nsight: Whether to profile Ray workers with nsight, see
             https://docs.ray.io/en/latest/ray-observability/user-guides/profiling.html#profiling-nsight-profiler.
+        placement_group: ray distributed model workers placement group.
         distributed_executor_backend: Backend to use for distributed model
             workers, either "ray" or "mp" (multiprocessing). If either
             pipeline_parallel_size or tensor_parallel_size is greater than 1,
diff --git a/vllm/engine/arg_utils.py b/vllm/engine/arg_utils.py
index bd44c2470182b..dab86b7c9eb35 100644
--- a/vllm/engine/arg_utils.py
+++ b/vllm/engine/arg_utils.py
@@ -548,14 +548,18 @@ def create_engine_config(self, ) -> EngineConfig:
                                    model_config.get_sliding_window(),
                                    self.enable_prefix_caching)
         parallel_config = ParallelConfig(
-            self.pipeline_parallel_size, self.tensor_parallel_size,
-            self.worker_use_ray, self.max_parallel_loading_workers,
+            self.pipeline_parallel_size,
+            self.tensor_parallel_size,
+            self.worker_use_ray,
+            self.max_parallel_loading_workers,
             self.disable_custom_all_reduce,
             TokenizerPoolConfig.create_config(
                 self.tokenizer_pool_size,
                 self.tokenizer_pool_type,
                 self.tokenizer_pool_extra_config,
-            ), self.ray_workers_use_nsight)
+            ),
+            self.ray_workers_use_nsight,
+            distributed_executor_backend=self.distributed_executor_backend)
 
         speculative_config = SpeculativeConfig.maybe_create_spec_config(
             target_model_config=model_config,