diff --git a/verl/workers/rollout/vllm_rollout/vllm_async_server.py b/verl/workers/rollout/vllm_rollout/vllm_async_server.py
index f4e26f13fde..4aea687fe66 100644
--- a/verl/workers/rollout/vllm_rollout/vllm_async_server.py
+++ b/verl/workers/rollout/vllm_rollout/vllm_async_server.py
@@ -412,11 +412,21 @@ async def run_server(self, args: argparse.Namespace):
             method="monkey_patch_model", kwargs={"vocab_size": len(self.model_config.tokenizer)}
         )
 
-        app = build_app(args)
-        if _VLLM_VERSION > version.parse("0.11.0"):
-            await init_app_state(engine_client, app.state, args)
+        build_app_sig = inspect.signature(build_app)
+        supported_tasks: tuple[Any, ...] = ()
+        if "supported_tasks" in build_app_sig.parameters:
+            supported_tasks = await engine_client.get_supported_tasks()
+            app = build_app(args, supported_tasks)
         else:
+            app = build_app(args)
+
+        init_app_sig = inspect.signature(init_app_state)
+        if "vllm_config" in init_app_sig.parameters:
             await init_app_state(engine_client, vllm_config, app.state, args)
+        elif "supported_tasks" in init_app_sig.parameters:
+            await init_app_state(engine_client, app.state, args, supported_tasks)
+        else:
+            await init_app_state(engine_client, app.state, args)
         if self.replica_rank == 0 and self.node_rank == 0:
             logger.info(f"Initializing a V1 LLM engine with config: {vllm_config}")