diff --git a/src/lighteval/models/vllm/vllm_model.py b/src/lighteval/models/vllm/vllm_model.py
index 8eadcf631..c144b6840 100644
--- a/src/lighteval/models/vllm/vllm_model.py
+++ b/src/lighteval/models/vllm/vllm_model.py
@@ -189,7 +189,6 @@ def _create_auto_model(self, config: VLLMModelConfig, env_config: EnvConfig) ->
         if int(config.data_parallel_size) > 1:
             self.model_args["distributed_executor_backend"] = "ray"
             self._batch_size = "auto"
-            return None
 
         model = LLM(**self.model_args)