diff --git a/vllm/executor/ray_gpu_executor.py b/vllm/executor/ray_gpu_executor.py
index 24b3a8c18d92..43cb37cfb5e0 100644
--- a/vllm/executor/ray_gpu_executor.py
+++ b/vllm/executor/ray_gpu_executor.py
@@ -154,6 +154,7 @@ def _init_workers_ray(self, placement_group: "PlacementGroup",
         scheduler_config = copy.deepcopy(self.scheduler_config)
         device_config = copy.deepcopy(self.device_config)
         lora_config = copy.deepcopy(self.lora_config)
+        vision_language_config = copy.deepcopy(self.vision_language_config)
         kv_cache_dtype = self.cache_config.cache_dtype
 
         # Initialize the actual workers with the Worker class.
@@ -172,6 +173,7 @@ def _init_workers_ray(self, placement_group: "PlacementGroup",
                     rank,
                     distributed_init_method,
                     lora_config=lora_config,
+                    vision_language_config=vision_language_config,
                     kv_cache_dtype=kv_cache_dtype,
                 ))