diff --git a/vlmeval/config.py b/vlmeval/config.py
index 80bf14dde..43c299c1d 100644
--- a/vlmeval/config.py
+++ b/vlmeval/config.py
@@ -333,6 +333,7 @@
     'Qwen-VL-Max-0809': partial(Qwen2VLAPI, model='qwen-vl-max-0809', min_pixels=1280*28*28, max_pixels=16384*28*28),
     'Qwen-VL-Plus-0809': partial(Qwen2VLAPI, model='qwen-vl-plus-0809', min_pixels=1280*28*28, max_pixels=16384*28*28),
     'Qwen2-VL-72B-Instruct': partial(Qwen2VLChat, model_path='Qwen/Qwen2-VL-72B-Instruct', min_pixels=1280*28*28, max_pixels=16384*28*28),
+    'QVQ-72B-Preview': partial(Qwen2VLChat, model_path='Qwen/QVQ-72B-Preview', min_pixels=1280*28*28, max_pixels=16384*28*28),
     'Qwen2-VL-7B-Instruct': partial(Qwen2VLChat, model_path='Qwen/Qwen2-VL-7B-Instruct', min_pixels=1280*28*28, max_pixels=16384*28*28),
     'Qwen2-VL-7B-Instruct-AWQ': partial(Qwen2VLChat, model_path='Qwen/Qwen2-VL-7B-Instruct-AWQ', min_pixels=1280*28*28, max_pixels=16384*28*28),
     'Qwen2-VL-7B-Instruct-GPTQ-Int4': partial(Qwen2VLChat, model_path='Qwen/Qwen2-VL-7B-Instruct-GPTQ-Int4', min_pixels=1280*28*28, max_pixels=16384*28*28),
diff --git a/vlmeval/vlm/qwen2_vl/model.py b/vlmeval/vlm/qwen2_vl/model.py
index 65629b0a9..4c931d7e8 100644
--- a/vlmeval/vlm/qwen2_vl/model.py
+++ b/vlmeval/vlm/qwen2_vl/model.py
@@ -88,7 +88,9 @@ def __init__(
             temperature=temperature,
             repetition_penalty=repetition_penalty,
         )
-        self.system_prompt = system_prompt
+        if system_prompt is None and 'qvq' in model_path.lower():
+            system_prompt = """You are a helpful and harmless assistant.
+        You are Qwen developed by Alibaba. You should think step-by-step."""
         self.verbose = verbose
         self.fps = 2.0
         self.nframe = 64