diff --git a/vllm/v1/worker/gpu_model_runner.py b/vllm/v1/worker/gpu_model_runner.py
index 97d8c91b4659..0190505b8a8f 100644
--- a/vllm/v1/worker/gpu_model_runner.py
+++ b/vllm/v1/worker/gpu_model_runner.py
@@ -338,11 +338,11 @@ def _update_states(self, scheduler_output: "SchedulerOutput") -> None:
         for new_req_data in scheduler_output.scheduled_new_reqs:
             req_id = new_req_data.req_id
             sampling_params = new_req_data.sampling_params
+            generator = torch.Generator(device=self.device)
             if sampling_params.sampling_type == SamplingType.RANDOM_SEED:
-                generator = torch.Generator(device=self.device)
                 generator.manual_seed(sampling_params.seed)
             else:
-                generator = None
+                generator.initial_seed()
 
             self.requests[req_id] = CachedRequestState(
                 req_id=req_id,